The Influence of Freedom and Choice in Action Selection and the Valence of Action-outcomes on the Sense of Agency

Transcription

1 Wilfrid Laurier University Scholars Laurier Theses and Dissertations (Comprehensive) 2016 The Influence of Freedom and Choice in Action Selection and the Valence of Action-outcomes on the Sense of Agency Zeynep Barlas Wilfrid Laurier University barl0270@mylaurier.ca Follow this and additional works at: Part of the Cognition and Perception Commons, Cognitive Psychology Commons, and the Theory and Philosophy Commons Recommended Citation Barlas, Zeynep Wilfrid Laurier University, "The Influence of Freedom and Choice in Action Selection and the Valence of Actionoutcomes on the Sense of Agency" (2016). Theses and Dissertations (Comprehensive) This Dissertation is brought to you for free and open access by Scholars Laurier. It has been accepted for inclusion in Theses and Dissertations (Comprehensive) by an authorized administrator of Scholars Laurier. For more information, please contact scholarscommons@wlu.ca.

2 The Influence of Freedom and Choice in Action Selection and the Valence of Action-outcomes on the Sense of Agency by Zeynep Barlas Submitted to the Department of Psychology in partial fulfilment of the requirements for Doctor of Philosophy in Psychology: Cognitive Neuroscience Wilfrid Laurier University Zeynep Barlas 2016

3 Declaration of Co-Authorship/Previous Publication Chapter 2 (Study 1): Barlas, Z., & Obhi, S. S. (2013). Freedom, choice, and the sense of agency. Frontiers in Human Neuroscience, 7(August), 514. doi: /fnhum Chapter 3 (Study 2): Barlas, Z., & Obhi, S. S. (2014). Cultural background influences implicit but not explicit sense of agency for the production of musical tones. Consciousness and Cognition, 28(1), doi: /j.concog Chapter 4 (Study 3): Under review as: Barlas, Z., Hockley, W. E., & Obhi, S. S. Effects of free action selection and pleasantness of action outcomes on the sense of agency. Chapter 5 (Study 4): Under review as: Barlas, Z., Hockley, W. E., & Obhi, S. S. Choice-level, outcome valence, and the sense of agency. Chapter 6 (Study 5): Under review as: Barlas, Z., Hockley, W. E., & Obhi, S. S. Freedom and fluency in action selection: Can freedom outweigh the influence of action selection fluency on the sense of agency? ii

4 Abstract Sense of agency (SoA) refers to the subjective experience that one is the author of their actions and the ensuing outcomes of these actions. Previous research have suggested that both sensorimotor processes and high level inferences can contribute to the SoA. In five experiments, the present thesis examined the effects of action selection processes and the valence of action-outcomes on the SoA. The majority of these experiments measured the SoA by obtaining both subjective feeling of control (FoC) judgments over the actionoutcomes, and assessing the size of intentional binding. Intentional binding refers to the perceived temporal attraction between actions and their outcomes, and has been suggested as an implicit measure of the SoA. Experiment 1 manipulated the number of action alternatives as low, medium, and high and examined the effect of choice-level on intentional binding. The results showed that binding was strongest when participants had the maximum number of alternatives, intermediate when they had medium choice-level, and lowest when they had no choice. Experiment 2 recruited western and non-western participants and focused on the impact of pleasantness of action outcomes on both intentional binding and FoC judgment. The results revealed that both western and nonwestern groups showed greater FoC ratings for the pleasant compared to unpleasant outcomes. Moreover, for the western group only, binding was stronger for pleasant compared to unpleasant outcomes. In Experiment 3, participants performed freely selected and instructed actions, which could produce pleasant or unpleasant outcomes. The results revealed stronger binding and higher FoC ratings in the free- compared to instructed-choice condition. Additionally, FoC ratings were higher for the pleasant compared to the unpleasant outcomes. Similarly, Experiment 4 varied the choice-level between one (instructed), two, three, and four alternatives while the outcome of any choice could be pleasant or unpleasant. The results showed that binding was stronger in the four-choice condition compared to one-, two-, and three-choice conditions, while FoC ratings were systematically increased as the choice-level varied from one to four, and were higher for pleasant compared to unpleasant outcomes. In Experiment 5, participants were primed with either action or neutral images and performed either free or instructed actions. Free actions could be preceded by either neutral (neutral-free) or action primes (primed-free), and instructed actions indicated performing either prime-compatible or prime-incompatible actions. The findings showed that both binding and FoC ratings indicated stronger SoA in the neutral-free condition compared to all remaining modes of action selection. Moreover, these two measures of the SoA were significantly correlated. The overall results from these studies indicate that situational factors surrounding actions determine the contribution of predictive, prospective, and retrospective mechanisms to intentional binding and subjective judgments of agency. Among these factors, the present thesis highlights that one s freedom in action selection and the availability of various action alternatives can strongly influence the SoA. iii

5 Acknowledgments I consider myself a most fortunate PhD candidate as I have always had tremendous support during my studies in Canada. To begin with, I would like to give many thanks to Wilfrid Laurier University and the department of Psychology for rewarding me the Ontario Trillium Scholarship, which has made my research possible. I am grateful to my primary advisor, Sukhvinder Obhi, who welcomed me into his lab before I moved to Canada. During my PhD, he has always encouraged me to brainstorm new ideas, embrace new challenges, and achieve success. He would readily lend support on all terms. I have also benefited tremendously from his experience in scientific thinking and writing as well as communicating ideas clearly. Over the last two years, I have also had the privilege to work with my co-advisor Bill Hockley, who has not only given me the opportunity to delve into memory research but has also been kind and supportive whenever I needed it. I also cannot thank Jeffery Jones enough. He has kept his lab open and welcoming and supported me during difficult times. I would also like to thank to Prof Shinobu Kitayama with whom I had the invaluable opportunity to engage in research on the cognitive neuroscience of cultural differences. And many thanks to Max Gwynn, Doreen Weise, Laurie Manwell, Todd Ferretti, Anneke Olthof, David Brown, and Roger Buehler for helping me improve my teaching skills. I am also grateful to our research assistants, May El-Bouri, Megan Grocholsky, and Shawna Davis, who have helped greatly with data collection. Throughout my PhD, I have also been blessed to have many loving and supportive colleagues and friends, May, Sarah, Adam, Nikki, Brittany, and Jeremy, just to name a few. I am so happy and fortunate to have met you all! I would not have come this far without the constant love and support of my family, who have always encouraged me to pursue my dreams. Endless thanks to my angel, Fatos, who has never given up on believing in me and standing by me unconditionally. And, Brent, thank you for all the love and care you have surrounded me with over the last year--and for returning my whining with flowers and joy. iv

6 Contents List of Tables... vi List of Figures... vii List of Abbreviations... x Chapter 1: General Introduction... 1 Chapter 2: Experiment Chapter 3: Experiment Chapter 4: Experiment Chapter 5: Experiment Chapter 6: Experiment Chapter 7: General Discussion Appendices References v

7 List of Tables Chapter 2 Table 2.1 Mean judgment errors in each condition 36 Chapter 3 Table 3.1 Demographic information for the western and non-western group 54 Table 3.2 Consonant and dissonant chords used in the study 55 Table 3.3 Mean judgment errors in each condition 60 Chapter 5 Table 5.1 Means and standard deviations of interval estimations in each choice-level, outcome valence, and actual delay condition 112 Table 5.2 Means and standard deviations of FoC ratings in each choice level, outcome valence, and actual delay condition 114 Chapter 7 Table 7.1 Summary of the designs and main results of Experiments vi

8 List of Figures Chapter 1 Figure 1.1 Demonstration of the intentional binding effect 10 Figure 1.2 The comparator model, adapted from Frith et al. (2002) 15 Chapter 2 Figure 2.3 Trial procedure in the operant condition 35 Figure 2.2 Mean perceptual shifts for key press and tone judgments 38 Figure 2.3 Mean overall binding as a function of action choice 40 Chapter 3 Figure 3.1 Illustration of a sample trial in the operant condition 58 Figure 3.2 Mean perceptual shifts in key press and chord judgments as a function of chord type for the western (A) and non-western (B) groups 62 Figure 3.3 Mean overall binding as a function of chord type for both western and nonwestern groups 63 Figure 3.4 Mean FoC ratings across both groups as a function of chord type 64 Figure 3.5 Mean liking ratings across both groups as a function of chord type 65 Chapter 4 Figure 4.1 Schematic illustration of the tasks completed by each group of participants (upper panel) and the sample trial procedure in the interval estimation and FoC rating tasks (lower panel) 81 Figure 4.2 Mean perceived intervals in free-choice and instructed choice conditions 85 Figure 4.3 Mean perceived intervals as a function of choice and delay 85 Figure 4.4 Mean FoC ratings as a function of choice and outcome valence 87 Figure 4.5 Mean FoC ratings as a function of delay 88 Figure 4.6 Mean RTs in the free-choice and instructed-choice conditions 89 Figure 4.7 Mean RTs in pressing each key 89 Figure 4.8 Correlation between the pleasant versus unpleasant difference scores of FoC and pleasantness ratings in the free-choice condition 92 vii

9 Figure 4.9 Correlation between the pleasant versus unpleasant difference scores of FoC and pleasantness ratings in the instructed-choice condition 92 Chapter 5 Figure 5.1 Schematic illustration of the tasks completed by each group of participants (upper panel) and the sample trial procedure in the interval estimation and FoC rating tasks demonstrated for each choice-level (lower panel) 109 Figure 5.2 Mean perceived intervals as a function of choice-level 112 Figure 5.3 Mean FoC ratings as a function of choice-level 114 Figure 5.4 Mean RTs as a function of choice-level 116 Figure 5.5 Mean RTs as a function of key 116 Figure 5.6 Mean effort ratings as a function choice-level 117 Chapter 6 Figure 6.1 Illustration of the procedure in the interval estimation and FoC rating tasks 136 Figure 6.2 Mean RTs as a function of selection mode and released finger 140 Figure 6.3 Mean effort ratings in each selection condition 141 Figure 6.4 Mean perceived action-outcome intervals in the interval estimation task for each selection condition 144 Figure 6.5 Mean perceived action-outcome intervals in the interval estimation task for each selection and actual delay condition 144 Figure 6.6 Mean FoC ratings for each selection mode 147 Figure 6.7 Mean FoC ratings for each selection and actual delay condition 147 Figure 6.8 Overall correlation between RTs and effort ratings demonstrated for one participant 149 Figure 6.9 Between subjects correlation between perceived intervals and effort ratings in the prime-incompatible condition 151 Figure 6.10 Between subjects correlation between FoC ratings and effort ratings in the prime-incompatible condition 152 Figure 6.11 Between groups correlations between perceived intervals and FoC ratings in the neutral-free, primed-free, prime-compatible, and prime-incompatible conditions 154 viii

10 Chapter 7 Figure 7.1 Schematic illustration of the processes through which freedom, choice-level, and outcome valence are suggested to influence the FoC judgments 179 Figure 7.2 Schematic illustration of the processes through which freedom, choice-level, and outcome valence were shown to influence binding 180 ix

11 ACC AG ANOVA BOLD FoC DLPC EBR EMG IPC IPL PET PFC RCZ RP RT SD SEM SMA SoA tdcs TMS TPJ List of Abbreviations Anterior Cingulate Cortex Angular Gyrus Analysis of Variance Blood-Oxygen-Level Dependent Feeling of Control Dorsolateral Prefrontal Cortex Eye Blink Rate Electromyography Inferior Parietal Cortex Inferior Parietal Lobe Positron Emission Tomography Prefrontal Cortex Rostral Cingulate Zone Readiness Potential Response Time Standard Deviation Standard Error of the Mean Supplementary Motor Area Sense of Agency Transcranial Direct Current Stimulation Transcranial Magnetic Stimulation Temporo-Parietal Junction x

12 CHAPTER 1 Chapter 1 General Introduction 1.1 The sense of agency Conceptualization of the SoA Measuring the SoA in experimental settings Explicit/Direct measures Implicit/Indirect measures Underlying mechanisms of the SoA The role of predictive processes The role of postdictive processes The interplay between predictive and postdictive processes Present dissertation: The influence of freedom and choice in action selection and the valence of action-outcomes on the SoA Preview of the experiments in this dissertation

13 CHAPTER The sense of agency Sense of agency (SoA) refers to the subjective experience that one is the author of their actions and the ensuing outcomes of these actions (Dewey & Knoblich, 2014; Gallagher, 2000; Haggard & Chambon, 2012; Haggard & Tsakiris, 2009). This experience is a crucial aspect of self-consciousness; it not only entails the distinction between one s self and others as the actor but also conveys the sense of having control over what one s actions change in the environment. When we switch on a light, for example, we unquestionably know that the lightening is changed by our pressing the switch. The SoA has important implications in morality and taking responsibility for one s actions (Haggard & Chambon, 2012; Haggard & Tsakiris, 2009), and it is closely linked to the notion of free will (Aarts & van den Bos, 2011; Haggard, Cartledge, Dafydd, & Oakley, 2004; Haggard, Clark, & Kalogeras, 2002; Haggard, 2008). Aside from the influence of SoA on responsibility and morality, it is imperative to understand the very nature of how we experience the SoA. Most of the time, the SoA in our daily routine of actions is so pervasive and diffused in ourselves that we do not reflect on our authorship of our actions or their consequences. In simple terms, we just know we are the actors who control external events occurring through our actions. There are times, however, that our agentic experience is distorted when we lose control over what action to take or when the consequences of our actions conflict with our intentions. The SoA, therefore, is a vulnerable phenomenon; it can be amenable to several factors and even fail to inform us of who is in control of the actions. This has been shown in both healthy individuals and several psychological and neurological disorders. For instance, individuals with no 2

14 CHAPTER 1 medical conditions can experience anomalous SoA when the source of actions or outcomes is ambiguous (Aarts, Custers, & Wegner, 2005; Haggard et al., 2004; Wegner & Wheatley, 1999). Moreover, disturbances of the SoA such as feeling a lack of control or misattributing agency have been observed in several disorders such as schizophrenia (Farrer et al., 2004; Frith, 2005; Haggard, Martin, Taylor-Clarke, Jeannerod, & Franck, 2003; Jeannerod, 2009; Kircher & Leube, 2003; Werner, Trapp, Wüstenberg, & Voss, 2014), motor conversion disorder (Kranick et al., 2013), obsessive compulsive disorder (Belayachi & Van der Linden, 2010; Belayachi & Van Der Linden, 2010), and anarchic hand syndrome (Blakemore, Wolpert, & Frith, 2002). Since the late 1990s, therefore, the quest to understand how the SoA comes about and what specific mechanisms are affected in the above-mentioned disorders have sparked great interest in both psychology and neuroscience domains of research. As is the case with other aspects of self-consciousness, the SoA is a difficult topic of study due to its subjective nature. Thus, the scientific investigation of the SoA requires careful establishment of the relevant concepts and theoretical frameworks as well as appropriate experimental designs and measures. The following sections provide a brief overview of these components pertaining to the examination of the SoA. The survey will begin with introducing the conceptualization of the SoA, which identifies the levels at which the SoA is experienced. The next section will then discuss how the SoA has been measured in experimental settings based on the changes in one s perception of their actions and the outcomes of these actions. These measures have been used extensively to test the accounts of the mechanisms of the SoA, which are presented in the following section. We shall see that the nature of the SoA is multi-faceted, and there are numerous factors 3

15 CHAPTER 1 that contribute the subjective experience of actions including motor planning and control mechanisms, prior thoughts, high level inferences, and various situational cues. The last section is devoted to the scope of the present dissertation, which mainly focuses on the role of freedom and choice level in action selection, and the nature of the consequences of actions. 1.2 Conceptualization of the SoA As noted before, we commonly experience the SoA in the form a tacit and unquestioned state of a phenomenon. According to the recently developed two-level account of the SoA (Bayne & Pacherie, 2007; Synofzik, Vosgerau, & Newen, 2008), this experience is described as the low level, non-conceptual, and pre-reflective SoA (Gallagher, 2000, 2007, 2011). At a higher level, the SoA is experienced through the reasoning that incorporates retrospective judgments and inferences. The high level SoA is thus conceptual and reflective in nature. Although the distinction between the low and high levels of the SoA has provided a conceptual framework, it has also raised questions regarding how to measure these levels and the potential differences between the underlying mechanisms. Regarding the measures of the SoA (see section 1.3), it was contended that the low level SoA could be indexed by implicit measures while explicit self-reports would quantify the higher level SoA. Furthermore, it was proposed that low level SoA emerges from sensorimotor processes that operate mainly prior to the movement by producing the motor commands and the anticipations of the consequences of the movement (see section , Blakemore et al., 2002; Frith, Blakemore, & Wolpert, 2000; Frith, 2005). The high level SoA, on the other hand, was suggested to rely on inferences drawn from the observation of actions 4

16 CHAPTER 1 and their outcomes as well situational cues (see section 1.4.2, Bayne & Pacherie, 2007; Moore, Middleton, Haggard, & Fletcher, 2012; Obhi & Hall, 2011; Synofzik et al., 2008; Wegner & Wheatley, 1999; Wegner, 2003, 2004). However, accumulating research to date has shown that the relationship between the low and high levels of the SoA and the sensorimotor versus inferential processes, respectively, might not be straightforward. Indeed, recent findings and theorizing suggest that, low-level and high-level agency can be influenced by sensorimotor or inferential processes. Before reviewing these processes and the relevant research, it is important to conceive the measures that are most commonly employed in the literature. 1.3 Measuring the SoA in experimental settings The methodologies of the relevant research have employed both explicit/direct and implicit/indirect procedures to measure the SoA. The former are concerned with conscious self-reports about subjective control and agency attribution. The implicit measures, on the other hand, rely on changes in subjective perception of the timing of actions and their outcomes as well as on the perceived intensity of sensory outcomes Explicit/Direct measures One way to measure the SoA is to directly obtain one s self-reflection on their sense of control or authorship. These explicit measures thus most commonly require participants to rate on a scale (e.g., a 10-point Likert scale) to indicate how much control they feel over action outcomes (e.g., Balslev, Cole, & Miall, 2007; Barlas & Obhi, 2014; Ebert & Wegner, 2010; Linser & Goschke, 2007; Metcalfe & Greene, 2007; Sato & Yasuda, 2005; Wenke, Fleming, & Haggard, 2010) or over their actions (e.g., Sebanz & Lackner, 2007; Wegner, Sparrow, & Winerman, 2004). Additionally, in the contexts 5

17 CHAPTER 1 where the source of the action-outcomes is rendered ambiguous, participants are asked to make direct judgments about the cause (i.e., me, computer, or a confederate) of the observed outcomes of actions (e.g., Aarts, Custers, & Marien, 2009; Aarts, Custers, & Wegner, 2005; Dijksterhuis, Preston, Wegner, & Aarts, 2008; Spengler, von Cramon, & Brass, 2009; Wegner & Wheatley, 1999) Implicit/Indirect measures Administration of the explicit measures based on self-report has been suggested to be highly prone to contamination by issues such as social desirability, impression management, and the limits of introspection on the part of participants (Metcalfe & Greene, 2007; Obhi, 2012; Schüür & Haggard, 2011). Alternative methodologies were then employed to include indirect measures to overcome these issues with the explicit measures. Two such indirect measures are sensory attenuation and intentional binding Sensory attenuation Sensory attenuation refers to reduced perception of the sensory outcomes produced by self-generated actions (Blakemore, Frith, & Wolpert, 1999; Blakemore, Wolpert, & Frith, 1998, 2000; Blakemore, 2003; Hughes, Desantis, & Waszak, 2013; Macerollo et al., 2015; Weiss, Herwig, & Schütz-Bosbach, 2011; Weiss & Schütz-Bosbach, 2012). Sensory attenuation is proposed to rely on the processes involved in motor preparation. More specifically, models of motor control system (Blakemore et al., 2002; Frith et al., 2000; Wolpert, Ghahramani, & Jordan, 1995; Wolpert, 1997) suggest that before the movement takes place, a copy of the motor command is sent to the so called forward model (see Section 1.4.1) which produces the predictions towards the sensory consequences of the movement. It is also proposed that these predictions are then 6

18 CHAPTER 1 compared to the actual outcomes of the movement, and sensory attenuation is suggested to result from the matching between predicted and actual outcomes (Blakemore et al., 1998, 2000; Frith et al., 2000). One common example is that self-tickling is experienced as less intense compared to being tickled by someone else (Blakemore et al., 1998, 2000). Sensory attenuation therefore enables the distinction between self and other generated actions, and is suggested to be a low-level sensory measure of the SoA (Synofzik et al., 2008). Sensory attenuation can be measured by its electrophysiological correlates or behaviorally by obtaining perceived intensity of sensory stimuli. The most prominent electrophysiological marker of sensory attenuation is the N1 potential, which is found to be reduced in response to a self-generated auditory stimulus relative to when the same stimulus is externally generated (Bäß, Jacobsen, & Schröger, 2008). Reduction in N1 amplitude has also been demonstrated when the stimuli are produced by voluntary compared to involuntary [e.g., Transcranial Magnetic Stimulation (TMS) induced] movements (Timm, SanMiguel, Keil, Schröger, & Schönwiesner, 2014). Additionally, sensory outcomes produced by their associated actions were shown to yield N1 attenuation more strongly than those that are incongruent with the actions (Hughes et al., 2013; Kühn et al., 2011). Importantly, studies with disorders have shown that N1 attenuation was reduced or absent in schizophrenia (e.g., Ford, Mathalon, Kalba, Marsh, & Pfefferbaum, 2001; Ford, Mathalon, Heinks, et al., 2001) and in patients with psychogenic movement disorder, who report having reduced or no control over their movements (Macerollo et al., 2015). 7

19 CHAPTER 1 Behavioral studies have provided further evidence that causal beliefs can influence the perceived loudness of outcome tones. It was found, for instance, that perceived loudness of the tones was reduced when participants believed that the tones were produced by their actions as opposed to by another person (Desantis, Weiss, Schütz- Bosbach, & Waszak, 2012). In sum, previous research suggests that sensory attenuation can be used as an implicit marker of the SoA and is prone to be influenced by both sensorimotor processes and causal beliefs Intentional binding Another implicit measure of the SoA relies on the perceived times of actions and their effects. In their seminal study, Haggard et al. (2002) measured the perceived times of actions and their outcomes while participants made voluntary key presses and passive (TMS induced to motor cortex) movements as they viewed a conventional clock (Libet, Gleason, Wright, & Pearl, 1983) on the screen. These key presses would sometimes produce a tone after 250 ms delay, and the association between key presses and tones was varied using baseline and operant conditions. In one block of the baseline condition (action-only), participants made a key press at a time of their choosing while fixating a rotating clock-hand on an on-screen clock. The key press did not produce any tone and participants judged the onset time of their key press by reporting the position of the clock-hand at the time of their movement. In a second block of the baseline condition (outcome-only), participants passively listened for a tone in each trial and judged the onset of the tone. In the operant conditions, the key presses would always produce a tone after a 250 ms delay and in separate blocks of trials, participants judged either the onset 8

20 CHAPTER 1 time of the key press or the tone. In this way, the authors could calculate the perceptual shifts in the times of key presses and tones across baseline and operant conditions (i.e., by calculating the difference between the judgment errors across these two conditions for key presses and tones). Interestingly, examination of these shifts showed that key presses in the operant voluntary condition were perceived to be occurring later (closer to the tone). In contrast, the onset of the tones was perceived earlier (closer to the key press). The perceived times of key presses and tones, therefore, were attracted towards each other in the voluntary actions (see Figure 1.1). However, this temporal attraction effect was not observed for the TMS induced passive movements. Haggard et al. (2002) suggested that perceived temporal attraction between actions and their outcomes was distinct to the voluntary actions. The authors thus coined the term intentional binding to refer to this effect. Since then intentional binding has received great attention of the relevant research to explore its relationship with the SoA. Although the clock paradigm has been used quite frequently, an alternative procedure to measure the intentional binding effect was also developed. This procedure involves obtaining the perceived duration estimates of the actions-outcome interval (Caspar, Cleeremans, & Haggard, 2015; Ebert & Wegner, 2010; Engbert, Wohlschläger, & Haggard, 2008; Moore & Haggard, 2010; Moore, Wegner, & Haggard, 2009; Obhi, Swiderski, & Farquhar, 2013). Notwithstanding the type of the procedure (i.e., the clock or the interval estimation paradigm), the purported link between intentional binding and the SoA is that greater perceptual shifts binding the times of actions and outcomes, as well as shorter estimations of the action-outcome interval, imply 9

21 CHAPTER 1 a stronger SoA (e.g., Engbert, Wohlschläger, Thomas, & Haggard, 2007; Ku, Brass, Haggard, & Kühn, 2012; Moore & Obhi, 2012; Wenke & Haggard, 2009). Baseline: Temporal judgments of individual events Outcome-only Action-only Operant: Temporal judgments of causal events Action Outcome Perceived times Action perceptual shift Intentional binding Figure 1.1 Demonstration of the intentional binding effect. In the baseline conditions, actions and outcomes occur independently. That is, in the action-only condition participants press a key which does not produce any outcome and they judge the onset time of their key press. In the outcome-only condition participants passively observe the outcomes (e.g., tones) and judge the onset of the outcomes. In the operant conditions, participants actions always produce the outcomes and they judge the onset times of either their actions or the outcomes of these actions. Perceptual shifts for actions and outcomes are calculated by subtracting the 10 Outcome perceptual shift

22 CHAPTER 1 judgment errors in the operant conditions from the corresponding baseline conditions for each judged event (e.g., action and outcome). In a typical binding effect, these shifts demonstrate that the perceived times of actions and their outcomes are shifted towards each other. Studies using these paradigms have further investigated whether intentional binding could indeed be specific to self-generated actions. In this regard, one line of evidence favors the intentional binding effect as a reliable measure of the self-agency. Haggard and Clark (2003), for example, compared the intentional binding effect between voluntary movements and movements that were intended but the execution of which was disrupted by TMS. Replicating the results of the previous study (Haggard et al., 2002), they observed the binding of actions and outcomes in voluntary movements. When these movements were disrupted by TMS, however, a reversal of intentional binding was found. That is, the perceived times of movements and resulting outcome tones were shifted away from each other (i.e., a repulsion effect). These results suggested that the intentional binding effect required not only the presence of intentions but also the successful execution of intended movements. The same repulsion effect was also observed, particularly on the perceived times of outcome tones, when participants themselves inhibited their intended actions (Haggard, Poonian, & Walsh, 2009). Further research provided evidence that intentional binding can be influenced by several agency related cues. Stronger binding of actions and outcomes was reported, for example, when one believed themselves as the source of the action-outcomes (Desantis, Roussel, & Waszak, 2011), with positive or pleasant compared to negative outcomes 11

23 CHAPTER 1 (Barlas & Obhi, 2014; Takahata et al., 2012; Yoshie & Haggard, 2013), and when the number of available action alternatives is at maximum within a specific context (Barlas & Obhi, 2013). Studies examining aberrant SoA in certain disorders have also reported that the intentional binding effect was reduced in high functioning autism spectrum disorder (Sperduti, Pieron, Leboyer, & Zalla, 2013) and motor conversion disorder (Kranick et al., 2013) while it was found to be exaggerated in schizophrenia (Synofzik & Voss, 2010; Voss et al., 2010). Finally, recent research on the brain mechanisms behind this effect suggests the involvement of pre-supplementary motor area (pre-sma). Accordingly, disrupting this area by theta-burst TMS (Moore, Ruge, Wenke, Rothwell, & Haggard, 2010b) and by transcranial direct current stimulation-tdcs (Cavazzana, Penolazzi, Begliomini, & Bisiacchi, 2015) have been shown to the reduce the intentional binding effect. Although these findings lend support to the view that intentional binding is strongly associated with the SoA, another line of evidence has challenged its specificity to selfgenerated actions. One counter notion is that intentional binding might simply reflect the perception of causality between two events (Buehner & Humphreys, 2009; Buehner, 2012). Another line of research have shown that the size of binding was indifferent between self-generated and observed actions (Moore, Teufel, Subramaniam, Davis, & Fletcher, 2013; Poonian & Cunnington, 2013; Wohlschläger, Haggard, Gesierich, & Prinz, 2003; but see Engbert et al., 2007). One interpretation of these findings is that there might be overlapping mechanisms through which agency is inferred in self- and other-generated actions (Moore et al., 2013; Poonian & Cunnington, 2013). 12

24 CHAPTER 1 The resolution of the debate whether intentional binding can merely be related to self-agency requires further investigation. At the moment, however, intentional binding remains a promising phenomenon in SoA research which could further elucidate its underlying processes and relationship with the SoA (see Moore & Obhi, 2012, for a review of intentional binding). 1.4 Underlying mechanisms of the SoA An important question probed by the relevant research is concerned with the underlying mechanisms that give rise to the SoA. On this line, two main streams of processes have been suggested to play important roles in influencing the subjective experience of agency. These processes are described under the terms of predictive (sensory-motor) and postdictive (inferential) accounts. Briefly put, the predictive account is based on the computational models of motor control mechanisms that are responsible for the acquisition and control of movements by calculating the sensory consequences of these movements. Importantly, the predictive account is heavily dependent on the processes that occur before the movement (Blakemore et al., 2002; Frith et al., 2000; Frith, 2005; Wolpert et al., 1995; Wolpert, 1997). The postdictive account, on the other hand, relies more strongly on the post-movement processes that operate on the perception of causality and inferences drawn upon the observation of both the movement and its outcomes (e.g., Wegner & Wheatley, 1999; Wegner & Sparrow, 2004; Wegner, 2004). Initially, these two streams of processes were viewed as (virtually) mutually exclusive in terms of how they address the underlying mechanisms of the SoA. Recent findings, however, have resulted in agreement that both streams of processes can influence the SoA and the degree of their contribution is determined on contextual and various other 13

25 CHAPTER 1 factors (Desantis, Weiss, et al., 2012; Moore & Fletcher, 2012; Moore, Wegner, et al., 2009; Synofzik, Vosgerau, & Lindner, 2009; Synofzik, Vosgerau, & Voss, 2013) The role of predictive processes As noted above, the predictive account of the SoA emerged out of popular computational models of motor control system (Blakemore et al., 2002; Frith et al., 2000; Wolpert et al., 1995; Wolpert, 1997). According to the so called comparator model (see Figure 1.2), motor learning and motor control are managed by the coupling of two main internal models, namely the inverse and forward models, and the three comparators that hold various functions. The major role of the inverse models is to issue the motor command to reach the desired state of the body in accordance with the goals of the agent. Once the motor command is issued, the efferent copy of this command is simultaneously sent to the forward models. The main function of the forward models is to predict the post-movement state of the body and the consequences of the movement. At this point, the role of the comparators becomes critical. The first comparator between the predicted state and the desired state informs the inverse models in case of a discrepancy so that any required adjustments to the motor planning can be performed before the movement takes place. The second comparator between the desired state and actual state, similarly, serves to tune the functioning of the inverse models for the improvement of motor learning of new actions. Finally, the third comparator detects the discrepancies between the predicted state and the actual state, and signals the outcome of this comparison back to the forward models. This comparison is crucial for two reasons. First, it helps the forward models improve their functioning in case of a mismatch between predictions and actual outcomes. Second, and more importantly, the result of this comparison is suggested to 14

26 CHAPTER 1 allow the distinction between self and other produced actions. More clearly, the greater the discrepancy between the predicted and actual states the more likely that agency is attributed to the others or the experience of self-agency is weakened. In support of the role of internal models on the SoA, Sato and Yasuda (2005) manipulated the congruency between actions (left and right key presses) and outcomes (high and low pitch tones) by rendering the outcomes unpredicted in terms of their timing and frequency. Their results showed the subjective ratings of being in control of the outcomes were reduced when the timing and the frequency of these outcomes were incongruent with the previously learned action-outcome associations. Furthermore, in a second experiment, they showed that participants could experience illusory sense of control over prediction-matching outcomes when in reality they were externally produced. These results suggested that the matching between predicted and actual outcomes could remarkably influence one s subjective feeling of control (FoC). In a similar vein, Linser and Goschke (2007) used subliminal priming of actioneffects and examined the influence of congruency between these primes and actual action-effects on participants subjective FoC over the outcomes. The results showed that the FoC was greater when the primed effects were compatible with the actual outcomes, suggesting that unconscious modulation of the internal predictions could influence the subjective FoC. Further support to the role of the predictive mechanisms came from a study which showed that deafferented patients failed to discriminate between self and other generated cursor movements on the screen in the absence of visual feedback,while the control group of healthy participants was better able to do so (Balslev et al., 2007). 15

27 CHAPTER 1 These results highlighted the contribution of the matching between predicted state of the body and the proprioceptive input. Goal Affordances Desired state Inverse models Motor command Forward models Predicted state Movement Actual state Sensory feedback Actual state Figure 1.2 The comparator model, adapted from Frith et al. (2002). Although the role of action control mechanisms and internal models appears to be indispensable in giving rise to the SoA, the predictive account cannot explain, for instance, the cases in which one can still experience some degree of authorship when the pre-movement predictions do not match the actual outcomes. The main drawback of the predictive account is that it does not, in its basic form, incorporate the contribution of other potential cues such as background beliefs, conscious judgments, and inferential 16

28 CHAPTER 1 processing linked to the SoA (for a critical review see, de Vignemont & Fourneret, 2004; Pacherie, 2008; Synofzik et al., 2008, 2013) The role of postdictive processes The postdictive (inferential) account is based on the Humean analysis on the perception of causality (Hume, 1888). Within this analysis, we perceive two events as causally related when they are temporally contingent and consistent. Along the same lines of causality and our interpretation of it, Wegner and Wheatley (1999) suggested that we infer that our actions or conscious thoughts are the cause of external events when (1) our thoughts or intentions occur before the observed events, (2) our intentions are consistent with the observed events, and (3) there are no other agents that could potentially cause the same events. As such, this account emphasizes the contribution of higher level inferences drawn retrospectively based on our observations of our actions and following events (see also, Wegner & Sparrow, 2004; Wegner, 2004). Support for this view came from the studies demonstrating that the SoA could occur even when the sensory-motor predictive signals are lacking. Wegner and Wheatley (1999), for example, conducted a study in which the participant and confederate simultaneously performed cursor movements on the screen. In some trials, the image on which the cursor would stop was presented through the headphones. When the primed image matched the actual image where the cursor was moved to, participants claimed authorship over these movements. Importantly, participants illusory judgment of agency occurred despite the fact that it was the confederate who actually caused the movements. In a similar vein, another study showed that participants could experience vicarious SoA over someone else s movements (Wegner et al., 2004). In this study, participants 17

29 CHAPTER 1 viewed their mirror reflection while the confederate stood behind the participant positioning their arms in the place of the participant s arms. In the mirror thus, it looked as if the confederate s arms belonged to the participant. Through headphones, movement instructions were delivered to both the participant and the confederate. Examination of the participant s judgments of how much control they felt over the mirror reflected movements revealed that the experience of agency was enhanced when the instructions and actual movements were the same. Taken together, these results supported the notion that the SoA is influenced by prior thoughts and situational cues. Nonetheless, the postdictive account too has its own inadequacies as a full-fledged account of the SoA. In essence, the main problem is that it merely emphasises post-movement cues and thoughts, leaving no role for internal premovement processes. It cannot explain thus, how in everyday life we experience the low level, pre-reflective SoA (see Section 1.2) without having to rely on high level judgments and inferences The interplay between predictive and postdictive processes The above mentioned drawbacks of predictive and postdictive accounts have led to the emergence of a new approach that combined the both processes into a unifying framework. This framework proposed a Bayesian model of cue integration process that estimates the weighting of both internal (sensory-motor, predictive) and external (postdictive, inferential) cues that are available in the context of actions. According to this model, the weighting of these cues determines their reliability and thus their differential influence on the SoA (Farrer, Valentin, & Hupé, 2013; Moore & Fletcher, 2012; Synofzik et al., 2013; Wolpe, Haggard, Siebner, & Rowe, 2013). 18

30 CHAPTER 1 Moore and Haggard (2008), for example, manipulated the probability of actioneffects to occur as high (75%) and low (50%), and measured the intentional binding effect using the clock paradigm. Participants made voluntary key presses and reported the onset time of either their key press or the auditory tone. The results showed that in the high probability condition, the perceived times of key presses were still shifted later in time when these key presses did not produce the tone. When the probability was low, on the other hand, perceptual shifts were still observed for the trials in which key presses produced the tone. The authors suggested that the former set of results pointed to the role of predictive processes while the latter indicated the contribution of retrospective inferences. In another study, participants were primed with auditory outcomes (low or high pitch) of voluntary and involuntary movements (Moore, Wegner, & Haggard, 2009). These primes could be either congruent or incongruent with the actual outcomes and participants estimated the temporal delay between their movements and auditory outcomes. It was found that binding was stronger in voluntary than involuntary movements and when the primes were congruent with the outcomes than they were incongruent. Importantly, however, the influence of prime compatibility was stronger in the involuntary condition compared to the voluntary condition. These findings indicated that when predictive cues are absent (as in the involuntary condition), external cues (i.e., the compatibility of the primes) can become more reliable and therefore influence the SoA (see also Stenner et al., 2014). Studies probing the neural structures associated with the SoA also support the multiple cue integration view. These studies suggest that a wide network of brain 19

31 CHAPTER 1 structures including the fronto-parietal network (Chambon, Wenke, Fleming, Prinz, & Haggard, 2013; Dogge, Hofman, Boersma, Dijkerman, & Aarts, 2014) is linked to the SoA. Among these areas, the cerebellum was proposed to be engaged in pre-motor predictions of the consequences of actions (Blakemore, Frith, & Wolpert, 2001; Blakemore et al., 1998) and dorsolateral prefrontal cortex (DLPC) was related to voluntary action initiation (Jahanshahi et al., 1995) and action monitoring (Rowe, Hughes, & Nimmo-Smith, 2010). Additionally, right inferior parietal lobe (IPL) was found to show increased activity when visual feedback is incongruent with the intended movement (David, 2010; David et al., 2007; Tsakiris, Longo, & Haggard, 2010). Accordingly, it was found that repetitive TMS applied over the inferior parietal cortex (IPC) resulted in rejection of self-agency for unperturbed feedback of the movements (Ritterband-Rosenbaum, Karabanov, Christensen, & Nielsen, 2014), suggesting that increased activity in this area signals the discrepancies between intended and actual outcomes of actions. Detection of such discrepancies and attribution of agency to the others was found to be particularly associated with angular gyrus (AG) in the right IPL (Farrer & Frith, 2002; Farrer et al., 2008; Miele, Wager, Mitchell, & Metcalfe, 2011; Spengler et al., 2009). In contrast, pre- SMA and rostral cingulate zone (RCZ) are suggested to mediate self-generated movements and action selection (Forstmann et al., 2008; Forstmann, Brass, Koch, & Cramon, 2006; Miele et al., 2011; Tsakiris et al., 2010). As noted before, the pre-sma is also suggested to be a major area underlying intentional binding (Cavazzana et al., 2015; Kühn et al., 2012; Moore et al., 2010b). Finally, conscious judgments of the degree of 20

32 CHAPTER 1 control over action-outcomes was linked to the increased activity in the prefrontal cortex (PFC) (Miele et al., 2011). 1.5 Present dissertation: The influence of freedom and choice in action selection and the valence of action-outcomes on the SoA Majority of the previous studies examining the role of predictive and postdictive processes have mainly focused on the compatibility of pre-motor predictions, prior thoughts and goals with the observed action-outcomes. However, it is also fundamental to human actions that performed actions are the result of a selection process among different actions. These processes can play important role in determining the right action to achieve one s intentions. In fact, several processes can be involved in the time course between the emergence of intentions and the outcomes of performed actions. In this regard, Pacherie (2008) proposed a comprehensive framework within which intentions are distinguished at three levels 1 based on their content and function to control and monitor human actions. According to this model, distal intentions (D-intentions) are represented at an abstract level and their realization to actions may occur within some flexible temporal delay (e.g., going to the park tomorrow). Proximal intentions (Pintentions), on the other hand, construct the representational plan of the action by integrating the conceptual information preserved in D-intentions and the current situational constraints (e.g., determining whether to walk or drive to go to the park based on time and weather constraints). Finally, motor intentions (M-intentions) are involved in 1 Pacherie (2008) established this framework by integrating previous views that distinguished between, for instance, prior intentions and intentions-in-action (Searle, 1983), future-directed and present-directed intentions (Bratman, 1987). 21

33 CHAPTER 1 the specification of the motor representations in terms of the spatial positions of the limbs to perform the action (e.g., different action plans would be programmed depending on whether one walks or drives to the park). It is important to note that both P-intentions and M-intentions can represent the selection of a specific action and the manner of action execution at different levels of specificity. Of more interest, this model also suggests that the subjective experience of agency is directly related to P-intentions and M-intentions as these levels are closely engaged in the monitoring and control of actions. The goal of the present thesis is twofold. First, it aims to uncover how the subjective experience of agency would be influenced when one s freedom in action selection (i.e., freely selected vs. instructed) and the number of available actions are manipulated. With respect to the dynamic theory of intentions (Pacherie, 2008) mentioned above, this manipulation specifically calls upon the levels of P-intentions and M-intentions which represent the selection and execution of actions. Second, the present thesis also aims to investigate the influence of perceived valence (e.g., pleasantness) of action-outcomes on the SoA. Importantly, these two manipulations (i.e., freedom and choice-level in action selection and the outcome-valence) were attempted to be implemented in both separate and common experimental contexts. Regarding the influence of externally perturbing the selection of actions on the SoA, Sebanz and Lackner (2007) found that FoC was reduced when external vocal instructions were incompatible with the stimulus guided actions. Under such external perturbations on action selection and execution, it was also found that participants felt stronger control when they could freely choose one of two actions compared to when they performed stimulus guided actions. These results suggested that external 22

34 CHAPTER 1 disturbances at the level of action selection (P-intentions) and action execution (Mintentions) can influence one s subjective experience of agency. A different line of research has suggested that action selection processes provide prospective cues to the SoA (Chambon et al., 2013). Relevant studies in this stream examined the role of action selection processes on the SoA along the lines of fluency (i.e., effortless processing of action selection) and the source (i.e., self vs. other) of action selection. More specifically, the role of selection fluency has been examined by using subliminal and supraliminal priming of actions. The particular goal of these studies was to investigate whether the compatibility of these primes with the actions would influence the SoA (Chambon & Haggard, 2012; Damen, van Baaren, & Dijksterhuis, 2014; Sidarus, Chambon, & Haggard, 2013; Wenke et al., 2010). Overall findings of this line of research suggested that compatible action primes, when subliminally presented, increased one s FoC over the action-outcomes (e.g., Chambon & Haggard, 2012; Wenke et al., 2010). The authors of these studies suggested that compatible primes could facilitate the selection processes, which in turn enhanced the sense of being in control of the outcomes produced by fluently selected actions (see Chambon, Sidarus, & Haggard, 2014 for a review). Studies investigating how the SoA could be influenced by self vs. other selected actions commonly included two alternative actions in their design. Wenke, Waszak, and Haggard (2009), for example, varied the timing and the choice of actions such that participants could either freely choose one of two keys or press the instructed key at a time of either their own choice or during a pre-specified interval. Across these conditions, the authors measured the size of intentional binding effect. Their results showed that the 23

35 CHAPTER 1 size of the binding between the perceived times of key presses and tones was greater when both the choice and timing of actions were specified by the same source, i.e., either freely selected or instructed, compared to when these dimensions were determined by different sources. The conclusion based on these results was that the SoA indexed by intentional binding could be enhanced when the decisive source of both the what- and the when-dimension of actions are the same as opposed to when different sources determine the timing and the type of actions. Regarding the second goal of the present thesis (i.e., the influence of outcome valence on the SoA), previous studies have shown that negative action outcomes (e.g. vocalization of fear) had an attenuating effect on intentional binding compared to positive (e.g., vocalization of amusement) or neutral (a pure tone) outcomes (Yoshie & Haggard, 2013) and positive monetary gains enhanced the binding effect. Also, priming participants with positive pictures compared to neutral ones was found to increase the intentional binding effect (Aarts et al., 2012). These results have commonly been interpreted with the notion of self-serving bias (Duval & Silvia, 2002; Miller & Ross, 1975; Taylor & Brown, 1994), which refers to the stronger tendency to attribute the self as the cause of positive than negative or undesirable events. In the present thesis, the role of action selection processes and the role of outcome valence were investigated in five experiments. The majority of these experiments measured the SoA using both the intentional binding paradigms and self-report measures of subjective control. In this way, we could observe the influence of these factors on both low SoA as purportedly indexed by intentional binding and high level of the SoA quantified by self-reports. 24

36 CHAPTER Preview of the experiments in this dissertation The goal of Experiment 1 was to examine how intentional binding is affected when the number of action alternatives is manipulated from low (no choice) to medium and high level of choice. In the no choice condition, participants could press the predetermined button on the response pad. In the medium-choice condition, they were free to choose among three buttons and in the high-choice condition, they were allowed to press any of the seven buttons. Participants reported the onset times of either the key presses or the outcome tones and we measured the size of binding in each condition. Experiment 2 examined the effect of outcome valence (i.e., pleasant vs. unpleasant) on the SoA. We recruited both western and non-western participants in order to explore any potential cultural differences in the effect of outcome valence. Participants completed a modified version of the intentional binding task in which they freely selected one of two keys, which produced either pleasant or unpleasant outcomes. We measured both intentional binding and FoC ratings over the outcomes. The goal of Experiment 3 was to investigate the influence of the origin of action selection (i.e., free vs. instructed) and the valence of the action-outcomes on both intentional binding and the FoC ratings. Participants performed either freely selected or externally instructed key presses among four options and each key press produced either a pleasant or an unpleasant auditory stimulus. Experiment 4 examined the influence of the number of action alternatives and the outcome valence on both intentional binding and FoC ratings. Participants were either free to choose a key among two, three, or four key alternatives or had only one (externally determined) option. Each key press could randomly produce either a pleasant 25

37 CHAPTER 1 or an unpleasant auditory stimulus. We also obtained the subjective ratings of mental effort experienced in key selection in each condition. The focus of Experiment 5 was the influence of freedom and fluency in action selection on the SoA. Accordingly, we used supraliminal action primes and participants performed either free or instructed actions in response to a symbolic target cue. The modes of action selection included free selections preceded by either neutral (neutralfree) or action primes (primed-free), and instructed selections required to perform either prime-compatible or prime-incompatible actions. All actions produced a tone after a jittered delay. Participants estimated the action-outcome delay and reported FoC judgments over the action-outcomes. Additionally, we obtained a subjective measure of perceived effort in action selection. 26

38 CHAPTER 2 Chapter 2 Experiment 1: Freedom, Choice, and the Sense of Agency Contents 2.1 Abstract Introduction Method Results Discussion Published as: Barlas, Z., & Obhi, S. S. (2013). Freedom, choice, and the sense of agency. Frontiers in Human Neuroscience, 7(August), 514. doi: /fnhum

39 CHAPTER Abstract The sense of agency (SoA) is an intriguing aspect of human consciousness and is commonly defined as the sense that one is the author of their own actions and their consequences. In the current study, we varied the number of action alternatives (one, three, seven) that participants could select from and determined the effects on intentional binding which is believed to index the low-level SoA. Participants made self-paced button presses while viewing a conventional Libet clock and reported the perceived onset time of either the button presses or consequent auditory tones. We found that the binding effect was strongest when participants had the maximum number of alternatives, intermediate when they had medium level of action choice and lowest when they had no choice. We interpret these results in relation to the potential link between agency and the freedom to choose one s actions. 28

40 CHAPTER Introduction One of the most fundamental aspects of human actions is the capacity to choose one s actions depending on the availability of a number of action alternatives (Haggard, 2008; Nichols, 2011). This capacity, however, is bound to the environmental circumstances that determine whether the environment offers a range of action alternatives and whether one can freely choose an action among these options or perform an action that is specified by external sources. The critical aspect of free actions is that the decisions that determine whether to act or not, what action to perform, and when to perform an action are self-generated (Brass & Haggard, 2008). Although a fine-grained scientific definition of self-generated actions (also referred to as voluntary, internally generated, or endogenous) is yet to be accomplished (for a discussion on this topic see Nachev & Husain, 2010; Obhi, 2012; Passingham, Bengtsson, & Lau, 2010a, 2010b; Schüür & Haggard, 2011), one approach is to consider it as in contrast to, for example, reflexes that are primarily stimulus driven actions. It is, however, important to note that this contrast does not mean that self-generated actions are completely independent from environmental sources (Filevich et al., 2013; Schüür & Haggard, 2011). When driving from one place to another, for example, the environment determines the possible routes, the distance to be travelled on each route, and the road conditions. In this case, one s decisions on the way to the destination would be dependent on these environmental constraints although one can still freely determine whether to drive or not, what route to take, and when to go. In this scenario, one s self involvement in making these decisions would be reduced if someone else, or an emergency situation, required one to travel at a 29

41 CHAPTER 2 specific time while one could still decide on what route to take. It would be even more reduced if an external source determined all decisions including which route to take. The critical point here is that freedom and self-generation of actions can be graded depending on the degree of self-involvement or endogenous processing (Passingham et al., 2010a, 2010b) in action selection, preparation, and execution. This is in line with the view that purports a continuum between self-generated actions and simple reflexes rather than placing the self-generated and externally influenced actions under two distinct categories (Haggard, 2008; Passingham et al., 2010a). Although the primary aim of this chapter, as well as the present thesis, is not to propose an extensive discussion on the conceptualization of self-generated actions, the term self-generated will be used to refer to relatively greater freedom and selfinvolvement (internal or endogenous processing) that the experimental context allows compared to more constrained conditions. As exemplified above, the degree of selfinvolvement can vary depending on who determines either the type or the timing of actions. The goal of the present chapter is to further unpack the what dimension of actions by varying the number of action alternatives and examine how the SoA as indexed by intentional binding would be influenced by this manipulation of choice-level in action selection. In this regard, the relationship between agency and freedom in action selection and the choice-level has been considered from various perspectives. From a broad perspective, agency and freedom are often considered to be tightly intertwined. More specifically, agency is thought to be strongest in an environment of opportunities (Pettit, 2001). Indeed, if a person cannot freely choose a course of action, the very notion that 30

42 CHAPTER 2 they are an autonomous agent is undermined. Given this, it might be expected that agency and freedom are related such that increasing levels of freedom to choose a course of action correspond to increasing levels of agency. This prediction is based on the abovementioned notion of greater self-involvement and internal processing in selfgenerated actions that are freely selected within some level of choice space. One relevant line of research examining the neural basis of free and instructed actions, for example, found increased BOLD (Blood-oxygen-level dependent) contrast in dorsolateral prefrontal cortex (DLPFC), inferior parietal lobe (IPL), rostral cingulate zone (RCZ), and supplementary motor area (SMA) when actions were freely selected as opposed to when they were performed as instructed (Cunnington, Windischberger, Deecke, & Moser, 2002; Filevich et al., 2013; Waszak et al., 2005). Among these areas, importantly, RCZ is suggested to be linked to free choice of varying number of action alternatives (Forstmann et al., 2008, 2006). Greater internal processing in free actions is also supported by the computational models of action selection. One such model, called the affordance competition hypothesis (Cisek, 2007), suggests that action selection relies on dynamic processing of representations of potential actions and sensory information related to the surrounding context. According to this model, critically, the representations of potential actions are in competition with each other to go under further processing during the course of action selection. Furthermore, it is suggested that the dorsal visual system involves in specifying the potential actions while the competition process among the representations of actions takes place in the fronto-parietal cortex. The competition of these representations consists of dynamic excitation and inhibition among the populations of neurons until one reaches 31

43 CHAPTER 2 a threshold activity strength. Additionally, this process also computes the sensory information received from prefrontal regions and basal ganglia. Therefore, this model confirms that self-generated actions are the outcome of the internal processes involving the agent s current needs, sensory information, and the representations of potential actions. The core idea of greater endogenous processing in free actions provides the theoretical grounds to examine how one s freedom to choose among a varying number of potential actions could influence the SoA. However, very few studies have addressed this question. Although it was not a direct examination of the link between freedom and agency, Wenke et al. (2010) assessed the subjective judgments of control when the compatibility between subliminal action primes and performed actions was manipulated in addition to varying the proportion of free versus cued trials. More specifically, participants could perform either freely selected (among two options) or cued (instructed) actions when the proportion of free trials was either high (75%) or low (25%). Additionally, subliminal action primes presented prior to the action selection could be either compatible or incompatible with the performed actions. The results showed that participants felt greater control over the outcomes when the primes were compatible with the performed actions, suggesting the effect of facilitating the action selection processes (see Chapter 6). Of more interest, the control ratings were higher when the proportion of free trials was high (75%) compared to when it was low (25%). This study suggested an intriguing link between one s freedom to choose an action and their feeling of control (FoC) over the consequences of their action. 32

44 CHAPTER 2 By extension, and reducing the general idea of a link between freedom and agency to a testable laboratory task, intentional binding might also be expected to vary with differences in the degree of freedom to choose an action. Again, agency and freedom are often talked about together and the feeling of freedom has been linked to choice (e.g., Markus & Schwartz, 2010). In this light, it is interesting to note that most previous intentional binding experiments have required participants to make a pre-specified action which is followed by a sensory event such as an auditory tone. In such cases, the participant is free to select when to make an action, but is not free to select which action to make. As proposed by Brass and Haggard (2008), decisions on which action to take (what), the timing of executing an action (when), and whether or not to execute an action (whether) are three important components of intentional actions(see also Haggard, 2008). By simply changing the number of action alternatives that are available to participants, it is possible to parametrically manipulate the environment of opportunities (i.e., choice) and thus ascertain the effect that the number of choice alternatives has on intentional binding. The fundamental question is, do more action alternatives produce greater levels of intentional binding than a more constrained choice set, where the agent is less involved in selecting which action to make? To this end, in the present study we examined how agency as purportedly indexed by intentional binding (e.g., Engbert, Wohlschläger, Thomas, & Haggard, 2007; Ku, Brass, Haggard, & Kühn, 2012; Moore & Obhi, 2012; Wenke & Haggard, 2009), is affected when the number of action alternatives is manipulated. To our knowledge, this is the first study that addresses the potential relationship between freedom of action choice and the SoA and intentional binding in particular. Accordingly, in the present study 33

45 CHAPTER 2 participants were requested to make a key press on a seven-button response pad while watching a conventional Libet clock on the screen. They reported their perceived times of key press or the auditory tone that was produced by their key press. In the no choice condition, they were told to press only one specific button on the response pad. In the medium-choice condition, they were free to choose among three buttons and in the highchoice condition they were allowed to press any of the seven buttons. For reports of the timing of actions and effects, we employed a similar paradigm to that of Libet, Gleason, Wright, and Pearl (1983) (see also Haggard et al., 2002; Obhi, Planetta, & Scantlebury, 2009). Based on the previously surveyed views regarding the relationship between freedom, choice, and the SoA as well as the emphasis on internal processing in free actions, we predicted that intentional binding would parametrically increase from the nochoice condition to medium-choice and to high-choice conditions. 2.3 Method Participants 24 right-handed participants (18 women; age range=17-22) took part in the study. All participants had normal or corrected-to-normal vision and received partial course credits for their participation. The study was approved by the Research Ethics Board of Wilfrid Laurier University, and all participants gave written informed consent prior to beginning the study. One participant s data was not included in the analyses due to not following the experimental instructions Apparatus and Procedure The experiment was programmed in Superlab 4.5 (Cedrus Corporation, USA) and ran on a Dell personal computer (3.07 GHz). The stimuli were presented on a 20 inch 34

46 CHAPTER 2 monitor (1600x1200). Participants sat approximately 60 cm away from the computer monitor and the responses were recorded on a laptop by the experimenter. The experiment consisted of baseline and operant conditions in which the number of keys to press (high: 7, medium: 3, no choice: 1) and the critical event (key press, tone) that participants judged the timing were manipulated. Similar to Haggard et al. s (2002) study, the baseline condition consisted of single events with either the key presses or the auditory tones. The key press single event condition included seven (high level of choice condition), three (medium level of choice condition), and one (no choice condition) key press choices. In the no choice condition, participants could only press the blue button centrally placed on the response pad. In the medium level of choice condition, they could choose any of the three buttons on the right side of the response pad. In the high level of choice condition, participants were free to choose any of the seven buttons on the response pad. When the critical event was the auditory tone, participants did not make any key press but only reported the time when they heard the tone. In the operant conditions, participants key press was followed by a 1000 Hz tone (duration: 100 ms, bit rate: 160 Kbps) presented after a delay of 200 ms and they were asked to report the time of either their key press or the tone. The condition (2: baseline, operant) together with the level of action choices (3: High, Medium, No choice), and the critical event (2: Key press, Tone) in total were tested in ten separate blocks with 30 trials each (see Table 2.1 for a list of different block types). The order of the blocks was randomized across participants. At the beginning of each block, participants were informed which key or keys they were allowed to press and which of the two events timing (key press or the tone) they were going to report. Participants completed 6 practice trials prior to the 35

47 CHAPTER 2 beginning of each block. Sixty practice trials in total thus were excluded from the data analysis. Each trial began with a warning signal noting that a new trial will begin, which remained on the screen for 1 s. The fixation cross was then presented for 500 ms and followed by the display of the Libet clock (1.8 cm in diameter) with a minute hand pointing to one of 12 positions marked at 5-minute intervals. Participants were told to report their judgments between 0 (12 o clock position) and 59, including the intermediate values. The minute hand remained stationary at the center of the screen for 500 ms and then started rotating clockwise at a 2.5 s period. In the baseline- where the single event was the key press only- and in the operant conditions, participants were told to make the key press at their own pace using their right index finger after the clock started rotating. They were instructed not to give stereotyped responses in the high and medium level of choice conditions and not to press the key at predetermined minute hand positions. In the baseline tone-only condition, participants did not make any key press but reported the onset of the tone occurred at a random time (jittered between 200 and 2000 ms) after the clock hand rotation started. The clock continued rotating for about 2000 ms after the participants reported the timing of the critical event. The perceptual times were verbally reported as minute hand positions and recorded by the experimenter on a laptop. At the end of the experiment, participants were debriefed and thanked for their participation in the study (See Figure 2.1 for a sample trial procedure). 36

48 CHAPTER 2 + Fixation, 500 ms Action Selection and Outcome High choice-level (seven buttons) Libet clock Medium choice-level (three buttons) No choice (one button) Time Key press Delay, 200 ms Consequent auditory tone 28 Participant s estimation of the key press or the tone onset in terms of the clock hand position Figure 2.1 Trial procedure in the operant condition. Each trial began with a fixation cross displayed for 500 ms. Participants then made a key press at their own pace after the clock started rotating. They were told to press a specific button in the no-choice condition or select one of three (medium level of choice) or seven (high level of choice) buttons on the response pad. The key press was followed by the auditory tone after a delay of 200 ms. In the baseline condition, participants either made a key press without hearing the tone and judged the timing of their key press, or heard the tone which occurred alone and judged the timing of the tone. 37

49 CHAPTER Results The experiment comprised a 2 (Condition: Baseline, Operant) x 3 (Level of choice: High, Medium, No choice) x 2 (Critical Event: Action, Tone) repeated measures design. After converting the clock hand judgments to time values in milliseconds, we calculated the judgment errors for each condition as the difference between perceived and actual times of events (Table 2.1). Trials with key press response times (RT) shorter than or equal to 500 ms and with judgment errors three standard deviations away from participant s average judgment error were excluded from the analysis. In addition, trials in which participants made a key press other than the permitted ones were removed from the data. The exclusion criteria resulted in the removal of 3.06% of all trials (range: 1-11%). Table 2.1. Mean judgment errors in each condition. For each event and each condition, perceived times were subtracted from the actual time of the corresponding events. Asterisks indicate the judged event (i.e., the onset time of key press or tone). Level of Choice Individual Event Mean Judgment Error SD No Choice Medium High Key press alone Key* tone Key tone* Key press alone Key* tone Key tone* Key press alone Key* tone Key tone* Tone alone

50 CHAPTER 2 We then obtained the perceptual shifts in terms of the difference between judgment errors between operant and the corresponding single event baseline conditions for both key press and tone judgments. For example, the perceptual shift for the high level action choice condition was calculated as the difference between the judgment errors in the operant-high-level condition from the baseline-high-level condition. Similarly, the perceptual shifts for the tone judgments were calculated as the difference between the judgment errors in each choice level-tone judgment condition and baseline-tone only condition. The positive shifts in the key press judgments and the negative shifts in the tone judgments relative to the corresponding baseline conditions demonstrate the temporal attraction, i.e. the intentional binding effect, between actions and effects (see Figure 2.2). We performed a 3 (Level of choice: High, Medium, No choice) x 2 (Critical event: Key press, Tone) repeated measures ANOVA to examine the effect of having different number of action choices on the perceptual shifts. The analysis revealed a significant main effect of key press choice (F(2,44)=3.36, p=.044, ƞ 2 =.13) and a significant main effect of critical event (F(1,22)=5.15, p=.003, ƞ 2 =.19). The interaction between these factors was also significant (F(2,44)=3.39, p=.043, ƞ 2 =.13). We predicted that binding would be least for the no choice condition, strongest for the high level of choice condition, and intermediate for the medium level condition. We thus conducted onetailed paired samples t tests to examine the two-way interaction in more detail. The t tests performed on the perceived times of actions revealed that when participants had high number of choices among which keys they could press, their perceptual shift in key press judgments from baseline condition was moved significantly 39

51 Perceptual shift from the baseline CHAPTER 2 further toward the tone compared to when they had medium level of choices (t(22)=2.29, p=.016) and to when they had no choice (t(22)=1.79, p=.043). The difference between medium level of choice condition and no choice condition was not significant (p>.05). With respect to the tone judgments, the perceptual shifts moved toward the perceived action onsets for both medium and high levels of choices. The size of the shift was greater for the medium level than the high level and it was in the opposite direction for the no choice condition. We found a significant difference in the perceptual shifts between high level of choice and no choice conditions (t(22)=-2.19, p=.020) and also between medium level of choice and no choice conditions (t(22)=-2.26, p=.017). The difference in the perceptual shifts between high and medium level of choices was not significant (p>.05) Mean perceptual shifts by choice level and judged event * * Key Press * * Tone High Medium Low/No-choice Figure 2.2 Mean perceptual shifts (difference between the judgment errors in the operant and baseline conditions) for key press and tone judgments (*p<.05). Error bars represent SEM. 40

52 Mean overall binding (ms) CHAPTER 2 We sought further the effect of choice levels on the mean overall binding which was calculated by subtracting the tone perceptual shift from the key press perceptual shift for each condition (Wenke, Waszak, & Haggard, 2009). We conducted a 3 (Level of choice: High, Medium, No choice) repeated measures ANOVA and found a significant main effect of action choice level on overall binding (F(2,44)=3.39, p=.043, ƞ 2 =.13). As expected, we found that overall binding was strongest in the high level of action choice condition, intermediate for the medium level of choice condition, and lowest for the no choice condition (see Figure 2.3). We performed one-tailed t tests to examine the differences across the three choice levels. The results showed that overall binding in the high level of choice condition was significantly greater compared to no choice (t(22)=1.99, p=.018) condition. However, the difference between high level of choice and medium level of choice condition as well as the difference between medium level of choice and no choice conditions were not significant (p>.05) Mean overall binding by choice level * High Medium Low Figure 2.3 Mean overall binding as a function of action choice. Error bars represent SEM (*p<.05). 41

53 CHAPTER Discussion Previous research focusing on different forms of the SoA has examined the contribution of various factors including predictive and retrospective processes (see Moore & Obhi, 2012 for a full review of these studies). Action selection is a crucial aspect of the agentic experience and has been shown to enhance the explicit feeling of control when facilitated by the subliminal priming of action alternatives (Wenke et al., 2010). The goal of the present study was to examine how intentional binding would be influenced by different levels of action choice. This is an important question given popular notions about how freedom and agency are intertwined (e.g., Pettit, 2001). We measured the perceived times of individual key press and tone events separately in both baseline and operant conditions which allowed us to compare the size of the perceptual shift between each level of action choice. First, we found that perceived times of key presses for all levels of choices were shifted forward in time. In the medium level and high level conditions, the direction of the perceived time of the tones was shifted toward the key press whereas, somewhat surprisingly, this was not the case for the no-choice condition. Importantly though, as Figure 2.2 shows, the overall shift for each individual event (i.e. key press and tone) were in the expected direction and demonstrate the intentional binding effect. Of more interest, we found that the degree of overall binding was greatest when participants had the highest level of action alternatives to choose from. In the medium choice condition, binding was not significantly different from the no choice condition, but both these conditions displayed less binding than the high choice condition. Moreover, the magnitude of the binding in three conditions displayed a parametric trend increasing from none to three and seven alternatives (see 42

54 CHAPTER 2 Figure 2.3). Thus, these results provide support for the notion that a high degree of choice is associated with greater action-effect binding than lower degrees of choice. These results serve to connect the SoA to free-choice and are also consistent with the common societal notion that the exercise of personal choice, freedom and agency are intimately intertwined (Hirschmann, 2003; Krause, 2012). What could be driving our observed effects of choice on intentional binding and by extension, the SoA? Given that all possible actions in the set of alternatives produced the same auditory event, our method could be construed as a true test of action selection on the SoA. That is, there is no obvious reason why an individual participant may have chosen one action over another, given that the outcome, or reward value of each possible action was fixed. Several explanations are possible. First, the results we report here are consistent with the finding that intentional binding is stronger when participants specify both the what and the when component of a pending action, compared to when they specify just one of these dimensions (i.e. when or what - (Brass & Haggard, 2008; Wenke et al., 2009). Participants in the present study were always responsible for specifying the when component, but had varying levels of choice about what action to make. Specifically, participants were constrained to just one possible action (no choice condition), three possible actions (medium choice condition) or seven possible actions (high choice condition). Thus, in the no choice condition, the action is completely specified externally by the experimenter whereas in both the medium and high choice conditions, the participant must internally specify which action they will ultimately select. By some accounts, the no choice condition can be thought of as more externally triggered than the medium and high choice conditions 43

55 CHAPTER 2 (see Obhi & Haggard, 2004; Obhi, 2012; Schüür & Haggard, 2011). Correspondingly, it has been shown that activation in areas associated with voluntary preparation to act, such as the supplementary motor area (SMA) is greater for actions that are more internally specified than externally specified (Jahanshahi et al., 1995). Thus one broad explanation for our findings is that more internal, endogenous processing prior to action production is linked to higher levels of agency experience, which manifests as greater intentional binding. Another interesting framework within which to consider the results of the present study is based on the affordance competition hypothesis that models behaviour as resulting from competition between different representations of potential actions (Cisek, 2007). In this model, action representations are thought of as distributed neural populations that are activated via selective attentional mechanisms (Tipper, Lortie, & Baylis, 1992). By such a view, the action that is finally selected and executed is chosen based on a dynamic reciprocal process operating largely within fronto-parietal circuits which involves mutual inhibition between potential action representations and is subject to biasing by excitatory inputs, some of which arise from cognitive decision making processes (see Cisek, 2007 for a detailed discussion). Within this framework, we suggest that high, medium, and no choice conditions differ in the degree of this dynamic activation and inhibition process that is ultimately responsible for action selection. Specifically, the no-choice condition may not involve the same degree of this dynamic inhibitory and excitatory activity as the high choice condition. We suggest that this difference might result in stronger activation of the 44

56 CHAPTER 2 representation of the action selected among many, such as in the high choice condition of the present experiment. This is akin to more endogenous processing being linked to greater agency, as suggested above, with the endogenous activity being specifically the dynamic interplay between excitatory and inhibitory processes during action selection. This explanation also predicts greater binding for the medium choice condition compared to the no choice condition as reported in our study, although the difference was not significant. From the present study, it appears that when seven alternative actions are available, this is sufficient to change the subjective experience of actions compared to when there is no alternative. However, three alternatives demonstrate no difference from seven or no alternatives. Clearly, more work is required to determine if this suggestion is tenable, but at the very least, our data do indicate that high choice affects binding in a way that no choice does not. One might argue that the cognitive load varied across three levels of action choices in our study, which could have contaminated our results. However, as previous studies discussed this concern in detail (e.g. Haggard et al., 2002), the errors in time judgments in the operant condition are subtracted from their corresponding baseline conditions (e.g. high level of choice action judgment errors in the baseline condition are subtracted from high level of choice action judgment errors in the operant condition) to calculate the perceptual shifts for each event and condition. Since the potential effect of either cognitive or attentional requirements varying across different levels of choice should be present in both baseline and operant conditions, this effect would diminish as a result of 45

57 CHAPTER 2 the subtraction we used to obtain the perceptual shifts. We thus feel confident in ruling out the effect of differential cognitive load across conditions. Having demonstrated that a high degree of choice is linked to increased binding, it is important to consider that there are limitations to the present study. For example, we did not assess the explicit SoA in this study and so cannot speak to how the number of action choice alternatives might affect the explicit feeling of agency. In addition, we did not manipulate the outcome of the different action alternatives. This is an obvious extension of the current work (see Chapter 4 and Chapter 5) and would allow for determining the influence of reward on intentional binding and the SoA. Despite these limitations, showing that intentional binding is influenced by the degree of action choice is an important finding and we believe the current study provides a new set of questions relating to how choice affects the SoA, which could apply to many domains that extend beyond a fundamental consideration of how the SoA arises. Finally, the current results bolster the notion that intentional binding is linked, in some complex way to agentic experience. It has previously been shown that priming low power reduces binding and activating memories of depression reduces binding (Obhi et al., 2013), whereas less versus more control of an aircraft, when control is shared with an automatic pilot, reduces binding (Berberian, Sarrazin, Le Blaye, & Haggard, 2012). Given that these scenarios are all accompanied by real changes in the degree of control that an individual either perceives themselves as having, or actually has, the idea that binding and agency are linked is strengthened. The key is for future work to understand why and precisely how the SoA and binding are affected by these kinds of manipulations. 46

58 CHAPTER 2 For now though, the current results reinforce the suggestion that increased personal choice increases agency which could form the foundation for a sense of freedom. In the following chapter, we turn our attention to the influence of the valence (i.e., pleasant vs. unpleasant) of action-outcomes on the SoA. After we examine how outcome pleasantness per se can influence the SoA in Chapter 3, we return back to the freedom and choice-level aspects of action selection and their link to the SoA. In Chapters 4-5, we advance the current design of study in such a way that free versus instructed actions (Chapter 4) or dynamically varying types of actions in different choice-levels (Chapter 5) can produce pleasant and unpleasant outcomes. 47

59 CHAPTER 3 Chapter 3 Experiment 2: Cultural Background Influences Implicit but not Explicit Sense of Agency for the Production of Musical Tones Contents 3.1 Abstract Introduction Method Results Discussion Published as: Barlas, Z., & Obhi, S. S. (2014). Cultural background influences implicit but not explicit sense of agency for the production of musical tones. Consciousness and Cognition, 28(1), doi: /j.concog

60 CHAPTER Abstract The sense of agency (SoA) is suggested to occur at both low and high levels by the involvement of sensorimotor processes and the contribution of retrospective inferences based on contextual cues. In the current study, we recruited western and non-western participants and examined the effect of pleasantness of action outcomes on both feeling of control (FoC) ratings and intentional binding which refers to the perceived compression of the temporal delay between actions and outcomes. We found that both western and non-western groups showed greater FoC ratings for the consonant (pleasant) compared to dissonant (unpleasant) outcomes. The intentional binding effect, on the other hand, was stronger for the consonant compared to dissonant outcomes in the western group only. We discuss the results in relation to how cultural background might differentially influence the effect of outcome pleasantness on low and high levels of the SoA. 49

61 CHAPTER Introduction In the previous chapter, we examined the effect of having a varying number of action alternatives on intentional binding. One of several other questions regarding agentic experience concerns situations where actions generate outcomes that differ in their valence or reward value. Indeed, most human actions are goal-directed and inextricably linked to the outcomes they produce (Elsner & Hommel, 2001; Elsner et al., 2002; Haggard, 2008; Herwig, Prinz, & Waszak, 2007). Previous research has examined how the reward value of action-outcomes can influence adaptive behavior and cognition in action control (e.g., Aarts, Custers, & Marien, 2008; Marien, Aarts, & Custers, 2012, 2013). This line of research has suggested that the reward signals related to the action-outcomes can increase the motivation and facilitate adaptive control of actions. Aarts et al. (2008), for example, showed that subliminally priming participants with words representing exertion paired with positive words increased the amount of effort displayed during squeezing a hand grip compared to priming with only exertion or positive words. These results suggested that positive primes could have acted as reward signals and thus enhanced the motivation to exert more effort in squeezing the hand grip. An intriguing question that results from these findings is whether and how the reward or positive value of action-outcomes would affect the subjective experience surrounding actions and the SoA. Previous studies in this vein have shown that negative action outcomes (e.g. vocalization of fear) had an attenuating effect on intentional binding compared to positive (e.g., vocalization of amusement) or neutral (a pure tone) outcomes (Yoshie & Haggard, 2013) and positive monetary gains enhanced the binding 50

62 CHAPTER 3 effect (Takahata et al., 2012). Also, priming participants with positive pictures compared to neutral ones was found to increase the intentional binding effect (Aarts et al., 2012). In their study, Aarts et al. (2012) presented neutral or positive pictures at the beginning of each trial and measured intentional binding using the clock paradigm (Haggard et al., 2002) and also the eye-blink rate (EBR) of the participants. The reason to include the EBR measurement in their design was to investigate whether potential influence of positive primes on binding could be mediated by EBR, which indirectly reflects the functioning of the dopaminergic system. Notably, previous studies showed that EBR was positively correlated with the concentration of dopamine (e.g., Taylor et al., 1999). Other studies have shown that dopamine agonists and antagonists had increasing and decreasing effects, respectively, on the EBR (e.g., Lawrence & Redmond, 1991). Moreover, dopaminergic system has long been known to involve in the processing of rewards (for a review see Ikemoto, Yang, & Tan, 2015) and in association of actions with their outcomes (Schultz, 2002). Given these findings regarding the link between dopamine and reward processing, action-outcome association, and EBR, Aarts et al. (2012) could examine if the potential effect of positive primes on binding could be explained by changes in EBR. Accordingly, their results showed that binding was stronger with positive than neutral primes. More interestingly, this effect was found to be moderated by individual differences in EBR. That is the difference in binding between positive and neutral primes was greater in individuals with higher spontaneous EBR compared to those with lower. Overall, these results suggested that the influence of positive valence on the SoA, as indexed by intentional binding, was mediated by the dopaminergic system. 51

63 CHAPTER 3 From the perspective of social psychology, the abovementioned findings indicating stronger binding with positive or rewarding outcomes can be interpreted with the notion of self-serving bias. Self-serving bias refers to that the tendency to attribute the self as the cause of outcomes is stronger for positive compared to negative or undesirable events (Duval & Silvia, 2002; Miller & Ross, 1975; Taylor & Brown, 1994). It has been argued, however, that there might be cultural differences in this bias. A recent meta-analysis of the relevant research examining cross-cultural differences in self-serving bias suggested that the self serving bias is stronger in U.S. and western than Asian samples (Mezulis, Abramson, Hyde, & Hankin, 2004). A critical extension to the abovementioned studies examining the relationship between the valence of action-outcomes and the SoA is whether perceived pleasantness can affect the SoA differentially based on potential cultural variations. Thus, in the present chapter we examined how intentional binding and the explicit FoC over action outcomes would be influenced when these outcomes differed in terms of their perceived pleasantness, which is potentially shaped by cultural differences. As action outcomes, we used consonant and dissonant piano chords that have long been subject to the study by researchers interested in music perception due to the different sensations they evoke in listeners. According to the Pythagorian view, the relative simplicity of the frequency ratio of two tones played simultaneously determines the pleasantness of the outcome sound (Helmholtz, 1877; Tenney, 1988). Consonance, in this regard, refers to the pleasantness produced by the co-occurrence of two tones whereas dissonance is described as unpleasant due to the beating and roughness (Dell Acqua, Sessa, Jolicoeur, & Robitaille, 2006; Dellacherie, Roy, Hugueville, Peretz, 52

64 CHAPTER 3 & Samson, 2011; Plantinga & Trehub, 2013; Shapira Lots & Stone, 2008). The major view regarding the perception of these tonal structures suggests that stability and pleasant-sounding attributes make consonance preferred over instable and roughsounding dissonance (Bidelman, Krishnan, & Bidelman, Gavin M.; Krishnan, 2009; McDermott & Hauser, 2004). However, the issue regarding the relationship between psychological and neurophysiological basis of consonance preference and its universal prevalence remains unresolved. One contention is that preference for consonance is innate and is due to certain constraints in the auditory system (Schellenberg & Trehub, 1996a; Tramo, Cariani, Delgutte, & Braida, 2001). In support of this view, studies with infants measuring their looking-time preference suggests that infants as young as 2 and 4-montholds (Trainor, Tsang, & Cheung, 2002) and 2-day-olds (Masataka, 2006) prefer to listen to consonant excerpts over dissonant ones. However, there is also accumulating evidence suggesting that consonant preference is the product of learning mechanisms. Vassilakis (2005), for example, examined Middle Eastern, North Indian, and Bosnian musical structures and noted that beats, which are thought to reside in dissonance, are well accepted in the musical structure of these cultures. In addition, Plantinga and Trehub (2013) tested consonance preference among 6-year-old infants and found that the listening time to the consonant chords was not longer than dissonant ones. Moreover, they showed that after a 3-minute exposure to either consonant or dissonant stimuli, infants listened to the familiar intervals longer, regardless of their consonant or dissonant status. 53

65 CHAPTER 3 The current study takes into account both lines of findings suggesting enhanced SoA over positive outcomes and cultural variances in the perceived pleasantness of consonance to address two important questions. First, these two types of stimuli would allow us to investigate whether low and high levels of the SoA are similarly affected by the pleasantness of action outcomes. Second, as consonance preference is suggested to vary across different cultures (e.g. Vassilakis, 2005), our design could reveal whether this variance can manifest itself on either low or high levels of the SoA. In the current study, participants completed a computer based task in which they made a voluntary right or left key press which was followed by either consonant or dissonant piano chords. We determined the intentional binding effect, subjective feelings of control (FoC) over the chords, and participants ratings for how much they liked each of consonant and dissonant chords. Based on the common bias towards attribution of the self as a cause of positive outcomes (Campbell & Sedikides, 1999), we predicted that the perceived pleasantness of consonant chords would produce higher FoC and liking ratings as well as stronger binding effect (Yoshie & Haggard, 2013) compared to the dissonant ones. As consonant and dissonant chords are based specifically on western tonal structure, our second prediction was that we would observe a greater effect of consonance in the western group compared to the non-western group. 3.3 Method Participants In total, 34 right-handed participants were recruited from the participant pool of Wilfrid Laurier University. The study was approved by the Research Ethics Board of 54

66 CHAPTER 3 Wilfrid Laurier University, and all participants gave written informed consent prior to beginning the study. We excluded four participants who, in at least one condition, had their mean judgment errors two standard deviations away from the sample mean. In addition, the data of one participant who could not follow the instructions were excluded from the analyses. Inclusion of these participants data was not found to affect the results reported below. We divided the remaining 29 participants into two groups based on the postexperimental questionnaire that gathered information about their cultural background. In this questionnaire, they indicated their country of origin and for how long they have been living in Canada. Additionally, they rated on two 10-point scales to indicate their lifetime level of exposure to and preference for western and non-western (i.e., Asian, African, and Middle East) music. For each participant, we calculated the index of exposure to western music by dividing the exposure rating for western music by the sum of ratings for western and non-western music. Similarly, we calculated the index of preference for western music over non-western music to examine the differences between the two groups (see Results). The western group included 17 participants (6 male, Mage = 21.5, SD= 5.2) who were born and raised in Canada, USA, or Western Europe. The non-western group consisted of 12 participants (5 male, Mage = 21.2, SD = 1.66) who were born in one of the non-western countries listed in Table 3.1. All participants had normal or corrected-tonormal vision and had no hearing problems. 23 of the participants received 11 CAD while the remaining group was granted with 1 course credit in return to their participation. 55

67 CHAPTER 3 Table 3.1 Demographic information for the western and non-western group. Group Western (n=17) Canada USA Western Europe Non-Western (n=12) Bosnia & Herzegovina China Hong Kong Iran Malaysia Pakistan United Arab Emirates Age Exposure rate for Western music Preference rate for Western music Number of years spent in Canada 21.5 (5.2).88 (.13).66 (.14) 18 (6.7) 21.2 (1.66).68 (.32).53 (.23) 7.6 (4.9) Apparatus and Procedure The experiment was programmed in Superlab 4.5 (Cedrus Corporation, USA) and ran on a Dell personal computer (3.07 GHz). Participants sat approximately 50 cm away from a 20 inch monitor (1600x1200) and the responses were recorded on a laptop by the experimenter. The auditory stimuli consisted of three consonant (perfect fifth, minor third, and perfect fourth) and three dissonant (minor second, major second, and tritone) piano chords and were recorded using Audacity All of the chords had the same 44.1 KHz sampling rate, 16 bit stereo format, and were 1.5 s in duration. The sound level of the chords was set to 80 db (See Table 3.2). 56

68 CHAPTER 3 Table 3.2 Consonant and dissonant chords used in the study. Chord Frequency Ratio Consonants Minor Third 6:5 Perfect Fourth 4:3 Perfect Fifth 3:2 Dissonants Minor Second 16:15 Major Second 9:8 Tritone 45:32 The first part of the experiment measured the effect of consonance status of the outcomes on intentional binding and consisted of two baseline and two operant conditions. Each of these four conditions was presented in randomly ordered blocks with 72 trials each. Each trial in the baseline-key press and operant conditions began with the screen indicating the start of a new trial (250 ms) which was followed by the fixation cross presented for 250 ms. The next display prompted the participants to choose either left or right button. Participants were free to choose either the left or the right button at a time of their choosing on a response pad using their right and left index fingers. They were told 57

69 CHAPTER 3 not to give any stereotyped responses when choosing right or left button and not to press the button at a predetermined time. The first key press then brought up the screen with a Libet clock on which the clock hand remained stationary for 500 ms and then started its rotation. Participants were told to press the same button at their own pace during the rotation. The reason why participants pressed the same key twice before and after rotation was to avoid any potential effect of the clock hand position (i.e. on the right or left half of the clock) biasing the participants right or left button choice. In the operant conditions, the second key press produced one of the six chords after a delay of 250 ms. For half of the participants, consonant and dissonant chords were produced by left and right button presses, respectively, and this matching was reversed for the remaining participants. In this way, left button press, for example, randomly produced one of the three consonant chords while the right button produced one of the three dissonant chords. The mapping of the key press and chord type was kept constant throughout the experiment for each participant. Depending on the critical event to be reported in a particular operant block, participants then judged the clock hand position (0 to 59) when either they pressed the button or when they first heard the chord (see Figure 3.1 (A)). In the baseline-key press condition the second key press did not produce any chord and participants judged the timing of their key press. The clock hand continued rotating for 2000 to 2500 ms after their verbal response regarding the time judgments and then the next trial began. In the baseline-outcome condition, each trial began with a warning signal followed by the fixation cross. The clock was then appeared and one of six different chords was randomly presented during the rotation. Participants judged the clock hand position when 58

70 CHAPTER 3 they first heard the chord. Time judgments were verbally reported and recorded on a laptop by the experimenter. After the intentional binding session was completed, participants performed another block of 72 trials to report their FoC over the chords (see Figure 3.1 (B)). Each trial in this block began with the message indicating the trial initiation (250 ms) followed by the fixation cross (250 ms). The next screen prompted the participants to freely choose one of the two buttons as in the intentional binding blocks. Their key press produced one of the six chords after a 250 ms delay and participants rated their FoC over the chord on a 10- point scale (1: not at all, 10: full control). In the last part of the experiment, participants passively heard each chord and rated on a 10-point scale to indicate how much they liked it (1: not at all, 10: very much). This block consisted of 18 trials in which all six chords were equally presented in a randomized order (see Figure 3.1 (C)). In total thus, participants completed five blocks with 72 trials each and one block with 18 trials throughout the experiment. After the experimental blocks, participants completed a demographic questionnaire which included items to note their origin of country, weekly amount of exposure to western and non-western music as well as their preference for each. Finally, they were debriefed and thanked for their participation. 59

71 CHAPTER 3 (A) Intentional Binding (B) Feeling of Control Judgment (C) Liking Judgment New Trial New Trial New Trial 250 ms 250 ms 250 ms ms 250 ms 250 ms When ready, press one of the two keys Until response When ready, press one of the two keys Until response Consonant 1500 ms Dissonant Consonant 1500 ms Dissonant Liking judgment for the chord Consonant 1500 ms Dissonant Feeling of control judgment for the chord Time judgment for either key press or the sound Figure 3.1 Illustration of a sample trial in the operant condition in the intentional binding (A), subjective FoC judgment (B), and liking judgment (C) sessions. 3.4 Results We excluded the trials with key press response times (RT) shorter than 600 ms and with time judgment errors being three standard deviations away from participant s average judgment error. In addition, trials in which participants made the second key 60

72 CHAPTER 3 press during the clock rotation different than the one on the previous step were removed from the data. The exclusion criteria resulted in the removal of 3.4 % of all trials (range: %) Musical Exposure and Preference We first compared the two groups in terms of their exposure to and preference for western music. Independent samples t test revealed that the exposure score was significantly higher in western (M=.87, SD=.13, N=17) than in non-western (M=.68, SD=.32, N=12) group, t(27)=2.17, p=.039, two-tailed. Similarly, the western group s ratings score for preferring western music (M=.69, SD=.14, N=17) was significantly higher than that of the non-western group (M=.53, SD=.23, N=12), t(27)=2.23, p=.034, two-tailed Button Choice We first examined whether the mapping between right/left button and consonant/dissonant outcome biased participants choice of key press. For each participant, we calculated the proportion of choosing right versus left button as well as the proportion of choosing the button that produced consonant chords. Paired samples t test revealed that participants chose the right button more often than the left one (M=52.80, t(28)= 3.19, p=.004). Although the ratio of choosing the button that produced consonant chords was higher than that produced dissonance (M= 50.40), the difference was not significant (p>.05) Intentional Binding In order to analyze the effect of consonance versus dissonance of action outcomes on intentional binding, we first obtained the perceptual shifts as the difference in the 61

73 CHAPTER 3 judgment errors between operant and the corresponding single event baseline conditions for both key press and chord judgments (see Table 3.3). Accordingly, the perceptual shifts for the key presses which produced consonant/dissonant chords were calculated as the difference between the judgment errors in the operant-consonant/dissonant (key press judgment) condition and the baseline (key press only) condition. Similarly, the perceptual shifts for the onset of the chord judgments were calculated as the difference between the judgment errors in the operant-consonant/dissonant (chord judgment) condition and baseline (chord only) condition. The positive shifts in the key press judgments and the negative shifts in the tone judgments relative to the corresponding baseline conditions demonstrate the temporal attraction, i.e. the intentional binding effect, between actions and outcomes (Haggard et al., 2002). Table 3.3 Mean judgment errors in each condition (C and D refers to consonant and dissonant, respectively. For the key presses, they refer to the associated the chord type). Condition Error (SD) Western Non-Western Baseline Key Press (C) (33.78) (48.96) Key Press (D) (33.20) (44.08) Chord (C) (35.36) (58.18) Chord (D) (45.29) (43.66) Operant Key Press (C) (68.91) (76.50) Key Press (D) (71.12) (77.18) Chord (C) (98.32) (108.96) Chord (D) (101.59) (110.55) 62

74 CHAPTER 3 We first conducted a 2 x 2 x 2 mixed-design, repeated measures ANOVA with chord (consonant, dissonant) and event (key press, chord) as the within subjects factors, and group (western, non-western) as the between subjects factor. The analysis yielded a main effect of event (F(1,27) = 90.16, p<.001, ƞ 2 =.77) suggesting that the perceptual shifts in key press and chord judgments were significantly different. Although we did not observe a main effect of chord on the perceptual shifts (p>.05), there was a significant three-way interaction between event, chord, and group (F(1,27) = 6.66, p=.016, ƞ 2 =.20). In order to examine the three-way interaction, we conducted 2 x 2 repeated measures ANOVA with chord (consonant, dissonant) and event (key press, chord) for each group (see Figure 3.2 (A) & (B)). For the western group, we found a significant main effect of event (F(1,16) = 53.04, p<.001, ƞ 2 =.77) as well as a significant interaction between chord and event (F(1,27) = 7.23, p=.016, ƞ 2 =.31). Paired samples t tests revealed that the difference in the perceptual shifts of key press judgments as well as the difference between the chord judgments for consonant and dissonant chords were not significant (all tests, p>.05). For the non-western group, the only main effect we observed was of event (F(1,11) = 39.9, p<.001, ƞ 2 =.78). 63

75 Perceptual Shift (ms) Perceptual Shift (ms) CHAPTER Mean perceptual shifts (A) Consonant Dissonant -150 Key Press Chord 100 Mean perceptual shifts (B) Consonant Dissonant -150 Key Press Chord Figure 3.2 Mean perceptual shifts in key press and chord judgments as a function of chord type for the western (A) and non-western (B) groups. Error bars represent SEM. Second, we calculated the overall binding by subtracting the tone perceptual shifts from the key press perceptual shifts (Wenke, Waszak, & Haggard, 2009). We then conducted a 2 x 2 mixed-design, repeated measures ANOVA with chord (consonant, 64

76 Overall Binding (ms) CHAPTER 3 dissonant) as the within subjects factor and group (western, non-western) as the between subjects factor. The test yielded a significant interaction between chord and group (F(1,27) = 6.66, p =.016, ƞ 2 =.20). We then examined the effect of chord type for each group separately and found that for the western group, the overall binding was significantly greater when the key presses produced consonant chords compared to dissonant ones (F(1,16) = 7.23, p=.016, ƞ 2 =.31). For the non-western group, however, the overall binding did not show difference between consonant and dissonant chords (all tests, p>.05, see Figure 3.3). Overall binding by group and chord * Consonant Dissonant Western non-western Figure 3.3 Mean overall binding as a function of chord type for both western and non-western groups. Error bars represent SEM (* indicates p<.05). Finally, we conducted linear regression analyses to explore if the time participants spent in Canada or the level of exposure to western music would predict the overall binding for consonant and dissonant chords. The dependent variables were overall binding for consonant chords and dissonant chords, and the difference in binding between 65

77 Mean FoC rating CHAPTER 3 consonant and dissonant chords. Number of years spent in Canada and the ratio of exposure to western music were simultaneously entered as the independent variables. None of the tests revealed any significant relationship between the level of familiarity with western music and intentional binding (all tests, p>.05) FoC Judgments In order to examine the effect of consonance versus dissonance on FoC judgments, we performed a 2 x 2 mixed-design, repeated measures ANOVA with chord (consonant, dissonant) as the within subjects factor and group (western, non-western) as the between subjects factor. The test revealed a main effect of chord (F(1,27) = 16.52, p<.001, ƞ 2 =.38) suggesting that participants felt significantly more in control over the consonant chords than the dissonant chords (see Figure 3.4). The interaction between the chord and group was not significant (p>.05). FoC rating by chord * Consonant Dissonant Figure 3.4 Mean FoC ratings across both groups as a function of chord type. Error bars represent SEM (* indicates p<.05). 66

78 Mean liking rating CHAPTER Liking Judgments A 2 x 2 mixed-design, repeated measures ANOVA with chord (consonant, dissonant) as the within subjects factor and group (western, non-western) as the between subjects factor yielded that participants liking ratings for consonant chords were significantly higher than dissonant ones (F(1,27)=63.70, p<.001, ƞ 2 =.70). There was no interaction between chord and group (p>.05). See Figure 3.5. Liking ratings * Consonant Dissonant Figure 3.5 Mean liking ratings across both groups as a function of chord type. Error bars represent SEM (* indicates p<.05). Finally, we calculated the difference between consonance and dissonance in overall binding, FoC ratings, and liking ratings. Bivariate correlation analysis showed that the difference score in the subjective control ratings significantly correlated with that in liking ratings (r (27) =.48, p<.001). 67

79 CHAPTER Discussion In the current study, we examined the effect of perceived pleasantness of the action outcomes on both intentional binding and subjective judgments of agency in western and non-western participants. We found that both groups felt significantly more control over the consonant chords than the dissonant ones and gave higher ratings of liking the former than the latter. The low level SoA indexed by the intentional binding effect was influenced by chord type in the western group only. That is, overall binding was significantly greater when western listeners actions produced consonant rather than dissonant chords whereas non-western listeners showed no differences in the binding effect between the two chord types. Another important result of the current study was that participants ratings for liking the consonant chords over the dissonant ones correlated with their respective FoC judgments. These results are noteworthy both in terms of consonance preference and cross cultural examination of the SoA at both low and high levels. Regarding consonance preference, both groups in our study reported liking consonant chords more than dissonant chords. Although the discussion about whether consonance preference is culture dependent or innate is beyond the scope of this paper, our results seem to support the notion of a universal preference for consonance. However, the group of non-western listeners in the current study was not completely isolated from exposure to western music. A cross-cultural comparison including a group with a completely different background of musical experience would provide a more solid ground to investigate this issue. For the moment, however, the explicit liking measure 68

80 CHAPTER 3 suggests a strong preference for consonance across individuals from various cultures (see also Fritz et al., 2009). Of more interest are the results regarding the effect of consonance status on the low and high level SoA and the differences between the two groups. Regarding the high level of the SoA as indexed by the FoC ratings in our study, the finding that both groups reported higher FoC over consonant than dissonant chords could strongly be related to the self-serving bias according to which causal attributions to self are stronger for positive than negative action outcomes (e.g., Campbell & Sedikides, 1999). The potential effect of self-serving bias on the FoC judgments becomes more tenable as we consider the finding that both groups reported to have found the consonant chords more pleasant than the dissonant ones. Moreover, this was positively correlated with the FoC ratings. It is thus fair to suggest that the difference we observed in agency judgments for two types of outcomes was driven by the self-serving bias. A more intriguing aspect of our findings concerns the differential effect of consonance status of action outcomes on intentional binding between two groups. To reiterate, we found that the western group showed greater binding for consonance than dissonance whereas the non-western group did not exhibit such an effect by the chord type. The crucial question here is why the western group showed stronger SoA over more pleasant outcomes at both low and high levels while the non-western group displayed the same effect only at the high level. If self-serving bias was the driving force for stronger agentic experience at the low level, we would expect both groups to display similar results on the intentional binding effect. However, previous studies provide deeper insight into how culture specific variations might influence the self serving bias and self 69

81 CHAPTER 3 evaluations in general. It has been suggested, for example, that the self serving bias is stronger in western than most of the Asian cultures (Mezulis et al., 2004). More importantly, cross cultural differences in the degree of self-evaluations and selfenhancement were found to be more apparent on implicit measures while explicit measures might not reveal any such difference (Hetts, Sakuma, & Pelham, 1999). Accordingly, Hetts et al. (1999) showed that Eastern immigrants showed conflicting results in associating self relevant prime words with positive or negative target words. That is, while the explicit measure of self-evaluation suggested that Eastern participants tend to associate the self concept more with the positive words just as American participants, response times taken as the implicit measure did show any bias towards selfenhancement in Eastern participants. On the basis of their results, the authors suggested that implicit measures reflect culture specific attitudes more readily than conscious evaluations of the self, which might be vulnerable to situational factors. Regarding our results, it is therefore fair to suggest that non-western participants showed a similar selfserving bias as the western group on the explicit judgments of agency while the two groups diverged in the effect of pleasantness of action outcomes on the low level SoA. In other words, relatively weaker bias of self-enhancement in non-western participants might have yielded no effect of outcome type on their low level SoA. An alternative explanation of our results concerns the potential difference in the degree of familiarity with consonance versus dissonance. It is possible that, for the western group who are more familiar with consonance than dissonance there is a difference in the quality of predictions produced by the forward model for consonant chords compared to the dissonant chords. Specifically, for the western participants, this 70

82 CHAPTER 3 difference in the quality of predictions or amount of motor preparation for consonant and dissonant chords might have yielded stronger overall binding between key presses and consonant chords. For the non-western group, on the other hand, the difference in the strength of predictions towards consonance versus dissonance might not be that obvious as they have presumably been exposed to the both in their cultural origin (Vassilakis, 2005). Taken together, our results suggest that pre-reflective and conscious experience of agency may be differentially affected by the cultural background of participants. This difference in the effects of cultural background on low level and high level agency supports the notion that the two forms of agency may be supported by dissociable neural mechanisms (Moore & Obhi, 2012). There are certain limitations to the current study that need to be addressed in future research. For example, for westerners we made an assumption that greater exposure to western music would imply a higher level of familiarity with and preference for consonant chords compared to dissonant chords. Similarly, we considered the nonwestern listeners reporting lower exposure to western music would bring about milder difference between consonant and dissonant chords. However, although the level of exposure might be a potential cause for the difference in how chord type affected the intentional binding in two groups, we did not find a significant relationship between the level of familiarity with western music and the binding effect. Further research should employ a more precise method to measure the level of exposure to western music by recruiting participants with a wider range of exposure from high to low. Another limitation of our study is that we did not measure our participants implicit or explicit 71

83 CHAPTER 3 status of self-serving bias as applied in previous studies which reported a conflict in the implicit and explicit belief systems in Eastern cultures (e.g. Hetts et al., 1999). Future experiments would provide a deeper insight if they employed valid measures of culture specific variations in the mechanism of self-evaluation. Finally, the baseline-chord condition in our study was different from the operant-chord condition in terms the predictability of the chord type and timing. However, we believe it is unlikely for this to contaminate our results as we did not find any effect of chord type on the chord judgment errors in the baseline and operant conditions. In sum, the current study raises several important ideas concerning the SoA and potential differences across cultures. First, we have shown that the perceived pleasantness of action outcomes influences the subjective judgement of the SoA such that more control is felt over desirable outcomes of actions. Second, the low level SoA indexed by the intentional binding effect can either parallel or not parallel the higher level judgment of agency depending on several possible factors, one of which appears to be cultural background and the level of prior exposure to consonant and dissonant tones. The current study also opens up a relatively new dimension of research concerning cross-cultural differences in the SoA. How culture interacts with the brain to shape an individual s phenomenological experience of their own actions is a fundamental question that we hope will spawn many interesting experiments in years to come. In the following chapter, the question of interest is how the SoA would be altered when actions are either freely chosen or performed as instructed and when these actions can produce either pleasant or unpleasant outcomes. 72

84 CHAPTER 4 Chapter 4 Experiment 3: Effects of Free Action Selection and Pleasantness of Action Outcomes on the Sense of Agency Contents 4.1 Abstract Introduction Method Results Discussion Under review as: Barlas, Z., Hockley, W. E., & Obhi, S. S. Effects of free action selection and pleasantness of action outcomes on the sense of agency. 73

85 CHAPTER Abstract Actions can be freely chosen or instructed and action outcomes can vary in pleasantness. To assess how these factors affect the sense of agency, participants performed freely selected or instructed key presses which could produce pleasant or unpleasant chords. We obtained estimates of the key press-chord intervals and feeling of control ratings (FoC) over the outcomes. Interval estimates were used to index intentional binding - the perceived temporal attraction between actions and their outcomes. Results showed stronger binding and higher FoC ratings in the free compared to instructed condition. Additionally, FoC was stronger for pleasant compared to unpleasant outcomes, and for pleasant outcomes that were produced by freely selected compared to instructed actions. These results highlight the importance of free action selection on the SoA. They also reveal how freedom of action selection and pleasantness of action outcome interact to affect the feeling of control. 74

86 CHAPTER Introduction The capacity to freely choose one s actions is fundamental to action control (Haggard, 2008; Nichols, 2011). Environmental conditions, however, can impose various factors that modulate one s freedom and self-involvement in action selection. As discussed in Chapter 2, the degree of self-involvement in actions can be varied by how much of the decisions regarding whether to act or not, what action to perform, and when to perform an action (Brass & Haggard, 2008; Haggard, 2008) is self-determined. In the study presented in Chapter 2, we manipulated the what dimension of actions such that the number of action alternatives could be either one, three, or seven. We reported that binding, as an indirect index of the SoA, was strongest when the context provided the highest number of alternatives (i.e., seven). Based on these results, we suggested that one s freedom to choose an action among (relatively) higher number of alternatives would bolster the SoA due to greater endogenous processing in the case of a large choice space. More clearly, selection of an action among high number of alternatives would result in greater activation of the final selection of an action compared to when one has none or few options. To reiterate, this interpretation was based on the affordance competition hypothesis (Cisek, 2007) that accounted for internal processing of action alternatives and suggested that an action is selective through the mechanism of dynamic inhibition and excitation of action representations. We also speculated that predictions produced by forward model (e.g., Blakemore, Wolpert, & Frith, 2002) towards the outcome of the selected action could also be stronger in the high-choice condition, which in turn could have led to greater binding compared to the no-choice 75

87 CHAPTER 4 condition. This study was the first to manipulate the choice-level in action selection and examine its impact on the SoA. Earlier research, in a similar vein, manipulated the source of at least one dimension of action decisions (i.e., whether, what, and when) as freely determined or externally instructed while the actions were limited to two alternatives. Wenke, Waszak, and Haggard (2009), for example, varied the timing and the choice of actions such that participants could either freely choose one of two keys or press the instructed key at a time of either their own choice or during a pre-specified interval. Using a similar paradigm to Haggard et al. (2002), participants were instructed to judge the time of either their key press or the resulting tone, in order to determine the size of the intentional binding effect across free and instructed choice conditions. Wenke et al. (2009) found that binding between the perceived times of key presses and tones was greater when both the choice and timing of actions were specified by the same source, (i.e., either freely selected or instructed), compared to when these dimensions were determined by different sources. On the basis of their results, the authors suggested that pronounced binding found in their study was owed to the compatibility of sources determining both the whatand the when-dimensions of actions. In their view, therefore, a conflict between the regarding sources would result in weaker binding. Another line of research investigated the neural basis of free versus instructed actions and have shown that the contrast between free choice and instructed actions was associated with increased BOLD activity in dorsolateral prefrontal cortex (DLPFC), inferior parietal lobe (IPL), rostral cingulate zone (RCZ), and supplementary motor area (SMA) (Cunnington et al., 2002; Filevich et al., 2013; Waszak et al., 2005). In an earlier 76

88 CHAPTER 4 study, the time of actions (i.e., extension of a finger) could be either self-initiated or externally triggered by the onset of auditory stimulus (Jahanshahi et al., 1995) and a PET (Positron Emission Tomography) scanning procedure was employed to measure the brain activity and the changes in movement related cortical potentials. Their results showed that self-initiated movements were associated with a specific network of brain areas including DLPFC, SMA, anterior cingulate, insular cortex, the lateral PMC, parietal area 40, the thalamus, and the putamen. Moreover, the peak amplitude of a movement related cortical potential, namely the readiness potential (RP), was greater in self-initiated compared to externally triggered movements. In another study, similarly, Obhi and Haggard (2004) assessed electromyographic (EMG) activity (reflecting the preparation of the muscles) in the right first dorsal interosseous while the onset time of participants finger press actions could be either selfinitiated or triggered by a tactile stimulus. The results showed that the EMG activity prior to action execution was greater when actions were self-initiated compared to when they were externally triggered. These results landed further support to the physical differences between self-initiated and externally triggered actions. Although the studies mentioned above, including the study in Chapter 2, attempted to understand the differences between free versus instructed actions on the basis of the underlying neural structures and the phenomenology of actions, questions remain whether these differences could also be salient depending on the value of actionoutcomes. As noted in Chapter 3, most human actions are goal-directed and related to the outcomes they produce (Elsner & Hommel, 2001; Elsner et al., 2002; Haggard, 2008; Herwig et al., 2007). In this regard, the reward or positive value of action-outcomes has 77

89 CHAPTER 4 been shown to enhance the motivational behaviour in actions (e.g., Aarts, Custers, & Marien, 2008) and also enhance the SoA indexed by intentional binding (Aarts et al., 2012; Takahata et al., 2012). In order to extend this line of research concerning the valence of outcomes, in Chapter 3, we examined the influence of pleasantness of outcome tones on both intentional binding and FoC ratings. Specifically, we used consonant and dissonant piano chords as outcome sounds that are- according to several physiological and psychological accounts of music perception- regarded as pleasant versus unpleasant, respectively (Dell Acqua, Sessa, Jolicoeur, & Robitaille, 2006; Dellacherie, Roy, Hugueville, Peretz, & Samson, 2011; Helmholtz, 1877; Plantinga & Trehub, 2013; Shapira Lots & Stone, 2008; Tenney, 1988; Bidelman & Krishnan, 2009; McDermott & Hauser, 2004; Schellenberg & Trehub, 2013; Tramo, Cariani, Delgutte, & Braida, 2001). To reiterate, the study of Chapter 3 assessed both FoC judgments and intentional binding while participants right or left key presses could produce either pleasant or unpleasant outcomes. We found that the amount of binding (in the western group only) and the subjective FoC over the chords was stronger when the outcome chords were pleasant compared to when they were unpleasant. These results supported the notion that positive or desired outcomes tend to be perceived as more strongly self-caused compared to negative, relatively undesirable outcomes. Moreover, this study promoted the investigation of cross-cultural differences in how agency (particularly at the low level) can be shaped by the valence of action-outcomes. To summarize, abovementioned findings demonstrate (i) activation differences in the brain between self-generated versus externally triggered actions (Cunnington et al., 78

90 CHAPTER ; Filevich et al., 2013; Forstmann et al., 2008, 2006; Jahanshahi et al., 1995; Waszak et al., 2005), (ii) the influence of source compatibility between the what and when dimension of action on binding (Wenke et al., 2009), (iii) stronger binding with high number of action alternatives (Chapter 2), and (iv) greater binding and FoC with pleasant compared to unpleasant outcomes (Chapter 3). One question, at this point, is to further probe how the SoA would be affected when the context includes both the manipulation of the source of action selection (free vs. instructed) and the valence of action-outcomes (pleasant vs. unpleasant). The goal of the present chapter is to address this question. Accordingly, participants performed either self-selected (free-choice) or externally specified (instructed) key presses that could randomly produce either a pleasant or an unpleasant chord. In the freechoice condition, participants could choose a key among four alternatives while in the instructed-choice condition, the selection was based on an instruction stimulus indicating which of the four keys to press. Between participants, we obtained estimations of the temporal interval between key presses and chords and FoC ratings over the chords. Based on the findings presented in Chapters 2-3, we expected stronger binding and higher FoC ratings in the free-choice than instructed-choice condition and when the outcome chords were pleasant than when they were unpleasant. 4.3 Method Participants In total, we recruited 46 undergraduate students from Wilfrid Laurier University. Participants were randomly assigned to either the interval estimation or the FoC rating task condition. Accordingly, 23 participants completed the interval estimation task (5 79

91 CHAPTER 4 male, 2 left-handed, Mage = 18.87, SD= 1.10) while the remaining 23 participants completed the FoC rating task (8 male, 5 left-handed, Mage = 19.17, SD= 1.77). All participants had normal or corrected-to-normal vision and had no hearing problems. The study was approved by the Research Ethics Board of Wilfrid Laurier University and participants gave written informed consent prior to beginning the study. Participants were compensated with course credits in exchange for their time Apparatus and stimuli The experiment was developed using Superlab 4.5 (Cedrus Corporation, USA) software and run on a Dell personal computer (3.07 GHz). Participants sat approximately 60 cm away from a 20 inch monitor (resolution: 1600x1200). Presentation of all stimuli was centered on a white background. Responses were made on a 5-key response pad. On this pad, four keys were placed on the right, left, up, and down side of the central key. An optical wheel mouse was used to indicate responses on visual analogue scales presented on the screen for interval estimation, FoC rating, and pleasantness rating tasks. The interval estimation scale was ranged from 1 to 1000 ms and marked at 50 ms intervals. FoC and pleasantness rating scales were marked at 0.5 point intervals from 1 to 6. Auditory stimuli consisted of two consonant (perfect fifth and perfect fourth) and two dissonant (minor second, major second) piano chords. These chords were recorded using Audacity 2.0.3, sampled at 44.1 khz with a 16 bit stereo format. Each chord was 1 s in duration and was presented at 60 db through the headphones Procedure A schematic of the tasks and the procedure is given in Figure 4.1. For each task of interval estimation and FoC rating, participants were first familiarized with the tasks and 80

92 CHAPTER 4 the stimuli, and completed 10 practice trials. Practice session was repeated only once for the participants who had difficulty in understanding the task and who made any errors in instructed-choice trials. Each task consisted of 288 trials in total, which were presented in a random order within 6 mixed blocks of 48 trials each. After completing each block, the experiment paused to allow participants to take a break, and continued after the experimental instructions for each task were presented on the screen. Each trial began with a 1 s presentation of an image representing the central key on the response pad. Participants were instructed to rest their left index finger on the central key when this image was presented. The following screen displayed one of five images representing either a specific key (right, left, up, down) or all four keys placed around the central key (see Figure 4.1). In the instructed-choice condition, only one specific key was presented and participants were required to press that exact key. In the free-choice condition, all four keys were presented and participants were free to choose any of the four keys. Participants were instructed to respond as fast as possible to the target stimulus and avoid giving stereotyped responses in the free-choice condition. The target stimulus remained on the screen until one of four keys was pressed. In case of an erroneous key press in the instructed-choice condition, a cross sign appeared on the screen and the trial ended. A valid response was followed by one of three delays (100 ms, 300 ms, 500 ms) before one of four auditory stimuli (1 s in duration) was presented. In the interval estimation task, participants were told that keypress-chord intervals would randomly vary between 1 and 1000 ms. After the chord was presented, the interval estimation scale was presented on the screen and participants were to indicate their estimation of the delay using the mouse with their right hand. No prior training was given for interval 81

93 CHAPTER 4 estimations and participants were told to rely merely on their sense of time when performing the interval estimations. In the FoC rating task, instead, the chord was followed by a 6-point visual analogue scale (1: the lowest level of control; 6: the highest level of control) participants were required to indicate the degree of control they felt over the production of the chord. They were told not to base their judgments on how fast or accurately they responded when making the key presses. Participants again used the mouse with their right hand and moved the cursor to any point on the scale and clicked to indicate FoC judgments. Inter-trial interval was set to 500 ms during which a blank screen was presented. 82

94 Time CHAPTER 4 Participants group #1 Interval estimation task Pleasantness rating task for the chords Participants group #2 FoC rating task Pleasantness rating task for the chords Free Instructed 1000 Until response Perfect fifth, Perfect fourth, Minor second, Major second Jittered delay 100 ms 300 ms 500 ms Interval estimation or FoC rating Jittered delay 100 ms 300 ms 500 ms Interval estimation or FoC rating Figure 4.1 Schematic illustration of the tasks completed by each group of participants (upper panel) and the sample trial procedure in the interval estimation and FoC rating tasks (lower panel). 83

95 CHAPTER 4 Each group of participants completing either the interval estimation or the FoC rating task finally performed the pleasantness rating task which aimed at measuring the subjective pleasantness of the chords used in the experiment. This task consisted of a block of 20 trials. Each chord was thus presented four times in a random order. The trials began with a 1500 ms presentation of the text message Listen. One of four chords was then delivered through the headphones and participants rated on a 6-point scale (1: very unpleasant; 6: very pleasant) to indicate how pleasant they found the chord. A 500 ms interval was placed before the next trial was presented. At the end of the experiment, participants were debriefed about the goal of the study and thanked for their time Data processing Raw data outlier exclusion For the interval estimation task, trials with RTs or interval estimations being three standard deviations away from the mean, or those with incorrect responses (pressing the wrong key in the instructed-choice condition) were excluded (Mexcluded = 2.18%, SD=.64% of all trials). The same criteria (except the interval estimation criterion) were also applied for the FoC rating task data (Mexcluded = 2.49%, SD=.73% of all trials) Participant exclusion The criteria to exclude a participant s data was having more than 20% of all trials excluded or failing to demonstrate a monotonic increase across the mean estimations of 100 ms, 300 ms, and 500 ms delays. No participant s data were excluded due to these criteria. 84

96 CHAPTER Data analyses A repeated measures analysis of variance (ANOVA) was conducted to examine the effects of choice (free, instructed) and valence (pleasant, unpleasant) on interval estimations and FoC ratings. RTs were analyzed as a function of key (right, left, up, down) and choice (free, instructed) while pleasantness ratings were analyzed by factoring in chord (perfect fifth, perfect fourth, minor second, major second). RTs and pleasantness ratings were analyzed combining the data from both interval estimation and FoC rating tasks. Greenhouse-Geisser correction was used where Mauchly s test of sphericity was violated. Post hoc multiple comparisons (Bonferroni corrected) were performed where differences across variable levels were examined. Additionally, two-tailed paired samples t-tests and one sample t-tests were conducted where appropriate. All data analyses were conducted using SPSS (version 16.0) and the significance level was set to Results Accuracy Mean percentages of accuracy in the instructed-choice condition was 99.34% (SD=.86) and 99.05% (SD=.98) in the interval estimation and FoC rating tasks, respectively Interval estimation We calculated the mean interval estimations for each level of choice, outcome valence, and delay. Accordingly, estimate data were subjected to a 2 x 2 x 3 repeated measures ANOVA with choice (free, instructed), valence (pleasant, unpleasant), and delay (100 ms, 300 ms, 500 ms) as within subjects factors. The analysis yielded a significant main effect of choice (F(1,22) = 5.71, p=.026, ƞ 2 =.21) such that interval 85

97 CHAPTER 4 estimations were significantly shorter when participants freely chose (M=420.59, SD=105.75) which key to press than when the key press was instructed (M=433.43, SD=108.77, see Figure 4.2). The main effect of delay was also significant (F(2,44) = , p<.001, ƞ 2 =.86), indicating that perceived intervals were significantly increased (p<.001, at all levels) across 100 ms (M=271.63, SD=90.50), 300 ms (M=428.37, SD=99.17), and 500 ms (M=581.03, SD=132.12). Outcome valence 2 did not have any significant effect or interactions with choice and delay on the perceived intervals (Fs<1, ps>.5). Finally, there was a significant interaction between choice and delay (F(2,44) = 4.20, p=.021, ƞ 2 =.16). In order to resolve the interaction, we performed paired samples t tests to compare the choice levels at each delay. Accordingly, the test revealed that perceived intervals at 100 ms were not significantly different between free (M=273.52, SD=84.99) and instructed (M=269.73, SD=86.59) conditions; t(22)=.48, p=.633, twotailed. At 300 ms, free choices yielded significantly shorter interval estimations (M=420.39, SD=94.68) compared to the instructed choices (M=436.34, SD=97.08); t(22)= -2.22, p=.037, two-tailed. Finally, at 500 ms, perceived intervals in the free condition (M=567.85, SD=126.84) were significantly shorter than the instructed condition (M=594.22, SD=129.52); t(22)= -2.85, p=.009, two-tailed (see Figure 4.3). 2 Although the number of trials for each condition is rather low (12), we also analyzed the influence of valence on the interval estimations by factoring in the chord type (perfect fifth, perfect fourth, minor second, major second) and yet did not find any significant effects. 86

98 Mean perceived interval(ms) Mean perceived interval (ms) CHAPTER 4 Perceived intervals by choice * Choice Free Instructed Figure 4.2 Mean perceived intervals in free-choice and instructed choice conditions (* p<.05). Error bars represent SEM Perceived intervals by choice and actual delay * * 100 ms 300 ms 500 ms Key press-chord delay Free Instructed Figure 4.3 Mean perceived intervals as a function of choice (free, instructed) and delay (100 ms, 300 ms, 500 ms) (* p<.05). Error bars represent SEM. 87

99 CHAPTER FoC ratings Mean FoC ratings were calculated for each choice type, outcome valence, and delay condition and were subjected to a 2 x 2 x 3 repeated measures ANOVA with choice (free, instructed), valence (pleasant, unpleasant), and delay (100 ms, 300 ms, 500 ms) as within subjects factors. The test revealed significant main effects of choice (F(1,22) = 8.03, p=.010, ƞ 2 =.27), valence (F(1,22) = 28.55, p<.001, ƞ 2 =.56), delay (F(2,44) = 9.09, p=.002, ƞ 2 =.29), and a significant interaction between choice and valence (F(1,22) = 8.61, p=.008, ƞ 2 =.28). No other significant effects or interactions were found by the analysis of FoC ratings (All Fs<1, ps>.6). More specifically, FoC ratings (see Figure 4.4) were significantly higher when choices were freely chosen (M=3.96, SD=.77) than instructed (M=3.79, SD=.67) and when outcome chords were pleasant (M=4.33, SD=.66) than they were unpleasant (M=3.43, SD=.78). Regarding the main effect of delay, post hoc tests showed that FoC ratings were systematically decreased (see Figure 4.5) as the delay increased from 100 ms (M=3.99, SD=.71), to 300 ms (M=3.85, SD=.72) and 500 ms (M=3.78, SD=.73). FoC ratings at 100 ms were significantly higher than both at 300 ms (p=.006) and 500 ms (p=.008). However, FoC ratings at 300 ms did not significantly differ from that at 500 ms (p>.4). Further analysis of the interaction between choice and valence revealed that FoC ratings were significantly higher over the pleasant outcomes when participants freely chose (M=4.46, SD=.67) which key to press than it was instructed (M=4.19, SD=.58); t(22)= 3.59, p=.002, two-tailed. However, the difference in the FoC ratings between free (M=3.46, SD=.83) and instructed (M=3.39, SD=.69) choices for the unpleasant outcomes 88

100 Mean FoC rating CHAPTER 4 was not significant; t(22)= 1.06, p=.302, two-tailed. Finally, for both free (Mpleasant=4.46, SDpleasant=.67; Munpleasant=3.46, SDunpleasant=.83) and instructed choices (Mpleasant=4.19, SDpleasant=.58; Munpleasant=3.39, SDunpleasant=.69) differences in the FoC ratings between pleasant and unpleasant outcomes were significant (t(22)= 5.28, p<.001; t(22)= 5.25, p<.001 for free and instructed choices, respectively) FoC ratings by choice and outcome valence ** Free * * Choice ** Instructed Pleasant Unpleasant Figure 4.4 Mean FoC ratings as a function of choice (free, instructed) and outcome valence (pleasant, unpleasant) (* p<.05, **p<.001). Error bars represent SEM. 89

101 Mean FoC rating CHAPTER FoC ratings by delay * * 100 ms 300 ms 500 ms Key press-chord delay Figure 4.5 Mean FoC ratings as a function of delay (100 ms, 300 ms, 500 ms) (* p<.05). Error bars represent SEM Response times (RTs) RTs (see Figures 4.6 & 4.7) were analyzed by a 2 x 4 repeated measures ANOVA with choice (free, instructed) and key (right, left, up, down) as within subjects factor. The test revealed a main effect of choice (F(1,45) = 22.68, p<.001, ƞ 2 =.33) such that choices were significantly slower in the free (M=636.32, SD=142.97) than instructed (M=586.71, SD=86.49) condition. The main effect of key was also significant (F(3,135) = 23.34, p<.001, ƞ 2 =.34). Post hoc tests revealed that pressing right (M=590.64, SD=107.26) and left (M=595.68, SD=107.02) keys were both significantly faster than pressing up (M=635.90, SD=125.85) and down (M=623.85, SD=118.77) keys (p right-up<.001, p rightdown<.001, p left-up<.001, p left-down=.001). Differences between right-left and up-down keys were not significant (all ps>.5). The interaction between choice and key was not significant (F(3,135) = 2.25, p=.107, ƞ 2 =.05). 90

102 Mean RT (ms) Mean RT (ms) CHAPTER 4 RTs by reponse ** Free 625 Instructed Response Figure 4.6 Mean RTs in the free-choice and instructed-choice conditions (** p<.001). Error bars represent SEM RTs by key ** ** ** * Right Left Up Down Key Figure 4.7 Mean RTs in pressing each key (* p<.05, ** p<.001). Error bars represent SEM. 91

103 CHAPTER Pleasantness ratings for the outcome chords As noted before, key press outcomes were one of two consonant (perfect fifth and perfect fourth) and two dissonant (minor second and major second) chords. We calculated the mean pleasantness ratings for each valence and ran paired samples t tests to compare the ratings. The test showed that consonant chords were perceived as more pleasant (M=4.51, SD=.64) than dissonant chords (M=2.48, SD=.42); t(1,45)=18.73, p<.001. We also conducted to a one-way repeated measures ANOVA with chord (perfect fifth, perfect fourth, minor second, major second) as a within subjects factor in order to examine differences across the four chords. The test revealed a significant main effect of chord (F(3,135) = , p<.001, ƞ 2 =.85). Post hoc tests indicated that perfect fifth (M=4.70, SD=.81) was perceived as more pleasant compared to perfect fourth (M=4.33, SD=.57, p=.001), major second (M=3.14, SD=.59, p<.001), and minor second (M=1.82, SD=.46, p<.001). Perfect fourth was also perceived as more pleasant compared to both minor second (p<.001) and major second (p<.001). Finally, minor second was perceived as more unpleasant compared to major second (p<.001). These results overall confirm that the consonant and dissonant chords we included in the experiment were indeed classified as pleasant and unpleasant action-outcomes Key selection in the free condition We also examined how the choice of key among for key alternatives in the free condition was distributed. Accordingly, the proportions of selecting right, left, up, and down keys were 28.79% (SD=11.61), 27.80% (SD=8.87), 21.27% (SD=7.27), and 22.13% (SD=8.51), respectively. A one-way repeated measures ANOVA with key (right, left, up, down) as within subjects factor revealed a main effect of key (F(3,135) = 6.01, 92

104 CHAPTER 4 p=.002, ƞ 2 =.85). Post hoc comparisons showed that the right key was selected more often than the up key (p=.019) and the left key was selected more often than the up key (p=.009). No other comparisons were significant (p>.05) Correlation Analyses FoC and pleasantness ratings In order to examine the relationship between FoC ratings and pleasantness ratings, we first calculated the difference in the mean FoC ratings in each choice (free, instructed) between pleasant and unpleasant outcomes (MFree(pleasant-unpleasant)=1.00, SDFree(pleasantunpleasant)=91; MInstructed(pleasant-unpleasant)=.80, SDInstructed(pleasant-unpleasant)=73). These differences were then subjected to bivariate Pearson correlation tests with the difference in the mean pleasantness ratings between pleasant and unpleasant outcomes (M(pleasantunpleasant)=2.25, SD(pleasant-unpleasant)=.73). The test revealed that the difference in the FoC ratings between the pleasant and unpleasant outcomes for both free (r=.50, p=.015) and instructed (r=.47, p=.024) conditions were significantly correlated with the difference in the pleasantness ratings, indicating that the more distant participants perceived the valence of the outcomes, the greater differences were felt in the FoC ratings between pleasant and unpleasant outcomes (see Figures 4.8 & 4.9). 93

105 CHAPTER 4 Figure 4.8 Correlation between the pleasant versus unpleasant difference scores of FoC and pleasantness ratings in the free-choice condition. Figure 4.9 Correlation between the pleasant versus unpleasant difference scores of FoC and pleasantness ratings in the instructed-choice condition. 94